The thing people miss in these “replace juniors with AI” takes is that juniors were never mainly about cheap hands on keyboards. They’re the only people in the org who are still allowed to ask “dumb” questions without losing face, and those questions are often the only signal you get that your abstractions are nonsense.
What AI does is remove a bunch of the humiliating, boring parts of being junior: hunting for the right API by cargo-culting Stack Overflow, grinding through boilerplate, getting stuck for hours on a missing import. If a half-decent model can collapse that search space for them, you get to spend more of their ramp time on “here’s how our system actually fits together” instead of “here’s how for-loops work in our house style”.
If you take that setup and then decide “cool, now we don’t need juniors at all”, you’re basically saying you want a company with no memory and no farm system – just an ever-shrinking ring of seniors arguing about strategy while no one actually grows into them.
"without losing face"? What culture are you referring to? The Western companies I have worked at do not discourage such questions -- in fact, it's often the sign of someone very senior when they ask a seemingly 'dumb' question that others have taken for granted.
Completely agree with this. I got to work closely with an IBM fellow one summer and I was impressed by his willingness to ask "dumb questions" in meetings. Sometimes he was just out of the loop but more often he was just questioning some of the assumptions that others in the room had left unquestioned.
Company culture. Some companies I worked for would fire you for questioning decisions. Others, welcomed criticism. You don’t really know which environment you’re in until someone says something. Are you going to take the risk and be the first?
This is a really good and under appreciated point. My recommendation to mid-level, senior, and staff engineers is to keep questioning decisions and create a culture where that’s encouraged.
Junior devs do that naturally (if you have the culture) because they don’t know anything already. It’s great
Usually the people who question decisions are shot down because they don’t have a wholistic understanding of the decision and (respectfully) don’t have good arguments. This is only because they are focused on some narrow aspect of the business which distorts or reduces their visibility and understanding.
I think you are right. It's still worth encouraging people to question decisions even though most of the time it won't be the right compromise for the business.
One of the most regulated industries, aviation requires crews to go through crew management training where it's explicitly trained for lower level staff to raise concerns in spite of perceived superior knowledge.
Some of the biggest accidents have happened directly due to this. Like Tenerife where the flight engineer had been listening to the radio and raised doubts about the runway being free but was ignored by the overconfident captain.
I don't know how to calculate "usually", in my experience people who question decisions in my company are shot down because the decision makers are usually (no pun intended) very incompetent and the questions make that visible, even if not intended. Many companies that I know are so corrupt that competent people are considered to be dangerous for the status quo.
> They’re the only people in the org who are still allowed to ask “dumb” questions without losing face
I strongly disagree, a Senior who cannot ask a "dumb question" is a useless developer to me. Ask away, we want to figure things out, we're not mind readers. Every good senior developer I know asks questions and admits when they don't know things. This is humility and in my eyes is a strong market for a good Senior Developer. The entire point of our job is to ask questions (to ourselves typically the most) and figure out the answers (the code).
Juniors could or should ask more questions, and by the time you're a Senior, you're asking key questions no matter how dumb they sound.
I think the biggest challenge now becomes how more seasoned engineers teach juniors. The AI makes the ramp a lot easier but you still do best when you understand the whole stack and make mistakes.
It’s damned near impossible to figure out where to spend your time wisely to correct an assumption a human made vs. an AI on a blended pull request. All of the learning that happens during PR review is at risk in this way and I’m not sure where we will get it back yet. (Outside of an AI telling you - which, to be fair, there are some good review bots out there)
Junior engineers now learn from AIs. And AIs now learn from RL cost functions. And RL cost functions are being set by PhDs, with little to no production grade engineering experience ;)
The result is interesting. First, juniors are miserable. What used to be a good experience coding and debugging, in a state of flow is now anxiously waiting if an AI could do it or not.
And senior devs are also miserable, getting apprentices used to be fun and profit, working with someone young is uplifting, and now it is gone.
The code quality is going down, Zen cycle interrupted, with the RL cost functions now at the top.
What do you mean by this?
I understand “cargo-culting” as building false idols, e.g. wooden headphones and runways to attract airplanes that never come.
It means to copy code or instructions from a site into your own project without having any comprehension of how or why it works.
example: you have a Windows problem. You search and read that "sfc /scannow" seems a popular answer to Windows problems. You run it without ever understanding what sfc does, whether the tool is relevant to your problem, etc. You are cargo culting a solution.
One is a "one junior per team" model. I endorse this for exactly the reasons you speak.
Another, as I recently saw, was a 70/30 model of juniors to seniors. You make your seniors as task delegators and put all implementation on the junior developers. This puts an "up or out" pressure and gives very little mentorship opportunities. if 70% of your engineers are under 4 years of experience, it can be a rough go.
That second model is basically the hospital model.
You have 1 veteran doctor overseeing 4 learning doctors. For example operating rooms do this, where they will have 4 operating rooms with 4 less experienced anesthesist and then 1 very experienced anesthesist who will rotate between the 4 and is on call for when shit hits the fan.
Honestly I think everyone here is missing the forest for the trees. Juniors their main purpose isn't to "ask questions", it's to turn into capable seniors.
That's also why the whole "slash our junior headcount by 3/4th" we are seeing across the industry is going to massively, massively backfire. AI / LLMs are going to hit a wall (well, they already hit it a while ago), suddenly every is scrambling for seniors but there are none because no one wanted to bear the 'burden' of training juniors to be seniors. You thought dev salaries are insane now? Wait until 4-5 years from now.
With the subjective view on what a junior is, I think the 70-30 - or higher - model is used in any company I ever interacted with. For this evaluation I consider junior=someone with less skills than needed to do the job autonomously/require direction and supervision most time, senior=someone that can work autonomously.
AI will replace jobs. People are putting their IT/dev budget into something, which means something else will be cut.
I also don’t believe juniors, kids, seniors, staff, principals, distinguished/fellow should be replaced by AI. I think they WILL be, but they shouldn’t be. AI at Gemini 3 Flash / Claude Opus 4.5 level is capable with help and review of doing a lot of what a lot of devs do currently. It can’t do everything and will fail, but if the business doesn’t care, they’ll cut jobs.
Don’t waste time trying to argue against AI to attempt to save your job. Just learn AI and do your job until you’re no longer needed to even manage it. Or, if you don’t want to, do something else.
This is actually a super power I have after spending my first part of my career in sales.
I was never formally trained so I just keep asking "why" until someone proves it all the way. Sales itself is also a lot about asking questions that won't come up to find the heart of the thing people actually want... which is just another side of the coin.
Yikes. Sounds like you work for a toxic company. The mid and senior level engineers I know all go out of their way to ask the dumb questions. Every junior employee I've mentored, I've told them the main way they can fail is not asking questions early and often. Gotta build a culture that supports questions.
What will eventually pan out is that senior devs will be replaced with junior devs powered by AI assistants. Simply because of the reasons you stated. They will ask the dumb important questions and then after a while, will even solve for them.
Now that their minds are free from routine and boilerplate work, they will start asking more 'whys' which will be very good for the organization overall.
Take any product - nearly 50% of the features are unused and it's a genuine engineering waste to maintain those features. A junior dev spending 3 months on the code base with Claude code will figure out these hidden unwanted features, cull them or ask them to be culled.
It'll take a while to navigate the hierarchy but they'll figure it out. The old guard will have no option but to move up or move out.
How do you suppose the old guard are filling their days now?
At some level, aren’t you describing the age-old process of maturing from junior to mid level to senior in most lines of work, and in most organizations? Isn’t that what advancing in responsibility boils down to: developing subtlety and wisdom and political credibility and organizational context? Learning where the rakes are?
I wish 3 months, or even 3 years, were long enough to fully understand the whys and wherefores and politics of the organizations I cross paths with, and the jungle of systems and code supporting all the kinds of work that happen inside…
Why would Claude code help you find unused features? The end customer uses features, not the AI. I would want to know from the end customer whether a feature is unused, and a Junior with an LLM assistant is not going to be able to tell me that without adding new features to the code base.
> The juniors working this way compress their ramp dramatically. Tasks that used to take days take hours. Not because the AI does the work, but because the AI collapses the search space. Instead of spending three hours figuring out which API to use, they spend twenty minutes evaluating options the AI surfaced. The time freed this way isn’t invested in another unprofitable feature, though, it’s invested in learning. [...]
> If you’re an engineering manager thinking about hiring: The junior bet has gotten better. Not because juniors have changed, but because the genie, used well, accelerates learning.
It may well be. Books have tons of useful expository material that you may not find in docs. A library has related books sitting in close proximity to one another. I don't know how many times I've gone to a library looking for one thing but ended up finding something much more interesting. Or to just go to the library with no end goal in mind...
Speaking as a junior, I’m happy to do this on my own (and do!).
Conversations like this are always well intentioned and friction truly is super useful to learning. But the ‘…’ in these conversations seems to always be implicating that we should inject friction.
There’s no need. I have peers who aren’t interested in learning at all. Adding friction to their process doesn’t force them to learn. Meanwhile adding friction to the process of my buddies who are avidly researching just sucks.
If your junior isn’t learning it likely has more to do with them just not being interested (which, hey, I get it) than some flaw in your process.
Start asking prospective hires what their favorite books are. It’s the easiest way to find folks who care.
I’ll also make the observation that the extra time spent is very valuable if your objective solely is learning, but often the Business™ needs require something working ASAP
Disagree, actually. Having spent a lot of time publishing papers in those very journals, I can tell you that just browsing a journal is much less conducive to discovering a new area to dive into than going to a library and reading a book. IME, books tend to synthesize and collect important results and present them in an understandable (pedagogical?!) way that most journals do not, especially considering that many papers (nowadays) are written primarily to build people's tenure packets and secure grant funding. Older papers aren't quite so bad this way (say, pre-2000).
When I first opened QBasic, <N> years ago, when I was a wee lad, the online QBasic help didn't replace my trusty qbasic book (it supplemented it, maybe), nor did it write the programs for me. It was just there, doing nothing, waiting for me to press F1.
I recall similar arguments being made against search engines: People who had built up a library of internal knowledge about where and how to find things didn't like that it had become so easy to search for resources.
The arguments were similar, too: What will you do if Google goes down? What if Google gives the wrong answer? What if you become dependent on Google? Yet I'm willing to bet that everyone reading this uses search engines as a tool to find what they need quickly on a daily basis.
I argue that there is a strong, strong benefit to reading the docs: you often pick up additional context and details that would be missing in a summary.
Microsoft docs are a really good example of this where just looking through the ToC on the left usually exposes me to some capability or feature of the tooling that 1) I was not previously aware of and 2) I was not explicitly searching for.
The point is that the path to a singular answer can often include discovery of unrelated insight along the way. When you only get the answer to what you are asking, you lose that process of organic discovery of the broader surface area of the tooling or platform you are operating in.
I would liken AI search/summaries to visiting only the well-known, touristy spots. Sure, you can get shuttled to that restaurant or that spot that everyone visits and posts on socials, but in traveling that way, you will miss all of the other amazing food, shops, and sights along the way that you might encounter by walking instead. Reading the docs is more like exploring the random nooks and crannies and finding experiences you weren't expecting and ultimately knowing more about the place you visited than if you had only visited the major tourist destinations.
As a senior-dev, I have generally a good idea of what to ask for because I have built many systems and learned many things along the way. A junior dev? They may not know what to ask for and therefore, may never discover those "detours" that would yield additional insights to tuck into the manifolds of their brains for future reference. For the junior dev, it's like the only trip they will experience is one where they just go to the well known tourist traps instead of exploring and discovering.
It's possible those arguments are correct. I wouldn't give up Google and SO, but I suspect I was learning faster when my first stop was K&R or a man page. There's a lot of benefit in building your own library of knowledge instead of cribbing from someone else's.
Of course no-one's stopping a junior from doing it the old way, but no-one's teaching them they can, either.
The naturally curious will remain naturally curious and be rewarded for it, everyone else will always take the shortest path offered to complete the task.
> The naturally curious will remain naturally curious and be rewarded for it
Maybe. The naturally curious will also typically be slower to arrive at a solution due to their curiosity and interest in making certain they have all the facts.
If everyone else is racing ahead, will the slowpokes be rewarded for their comprehension or punished for their poor metrics?
The thing is you need both. You need to have periods where you are reading through the docs learning random things and just expanding your knowledge, but the time to do that is not when you are trying to work out how to get a string into the right byte format and saved in the database as a blob (or whatever it is). Documentation has always has lots of different uses and the one that gets you answers to direct questions has improved a bit but its not really reliable yet so you are still going to have to check it.
No, trying stuff out is the valuable process. How I search for information changed (dramatically) in the last 20 years I've been programming. My intuition about how programs work is still relevant - you'll still see graybeards saying "there's a paper from 70s talking about that" for every "new" fad in programming, and they are usually right.
So if AI gets you iterating faster and testing your assumptions/hypothesis I would say that's a net win. If you're just begging it to solve the problem for you with different wording - then yeah you are reducing yourself to a shitty LLM proxy.
Not really. There’s a pattern to reading docs, just like there’s a pattern to reading code. Once you grasped it, your speed increase a lot. The slowness that junior has is a lack of understanding.
Complaining about docs is like complaining about why research article is not written like elementary school textbooks.
Disagree. While documentation is often out of date, the threshold for maintaining it properly has been lowered, so your team should be doing everything it can to surface effective docs to devs and AIs looking for them. This, in turn, also lowers the barrier to writing good docs since your team's exposure to good docs increases.
If you read great books all the time, you will find yourself more skilled at identifying good versus bad writing.
It really depends on what's being learned. For example, take writing scripts based on the AWS SDK. The APIs documentation is gigantic (and poorly designed, as it takes ages to load the documentation of each entry), and one uses only a tiny fraction of the APIs. I don't find "learning to find the right APIs" a valuable knowledge; rather, I find "learning to design a (small) program/script starting from a basic example" valuable, since I waste less time in menial tasks (ie. textual search).
Struggling with poorly organized docs seems entirely like incidental complexity to me. Good learning resources can be both faster and better pedagogically. (How good today's LLM-based chat tools are is a totally separate question.)
Nobody said anything about poorly organized docs. Reading well structured and organized complex material is immensely difficult. Anyone who’s read Hegel can attest to that.
And yet I wouldn’t trust a single word coming out of the mouth of someone who couldn’t understand Hegel so they read an AI summary instead.
There is value in struggling through difficult things.
Feel free to waste your time sifting through a dozen wrong answers. Meanwhile the rest of us can get the answers, absorb the right information quickly then move on to solving more problems.
And you will have learned nothing in the process. Congratulations, you are now behind your peer who "wasted his time" but actually knows stuff which he can lean on in the future.
1965: learning how to punch your own punch cards is part of the learning process
1995: struggling with docs and learning how and where to find the answers part of the learning process
2005: struggling with stackoverflow and learning how to find answers to questions that others have asked before quickly is part of the learning process
2015: using search to find answers is part of the learning process
2025: using AI to get answers is part of the learning process
To the extent that learning to punch your own punch cards was useful, it was because you needed to understand the kinds of failures that would occur if the punch cards weren't punched properly. However, this was never really a big part of programming, and often it was off-loaded to people other than the programmers.
In 1995, most of the struggling with the docs was because the docs were of poor quality. Some people did publish decent documentation, either in books or digitally. The Microsoft KB articles were helpfully available on CD-ROM, for those without an internet connection, and were quite easy to reference.
Stack Overflow did not exist in 2005, and it was very much born from an environment in which search engines were in use. You could swap your 2005 and 2015 entries, and it would be more accurate.
> To the extent that learning to punch your own punch cards was useful, it was because you needed to understand the kinds of failures that would occur if the punch cards weren't punched properly. However, this was never really a big part of programming, and often it was off-loaded to people other than the programmers.
I thought all computer scientists heard about Dijkstra making this claim at one time in their careers. I guess I was wrong? Here is the context:
> A famous computer scientist, Edsger Dijkstra, did complain about interactive terminals, essentially favoring the disciplined approach required by punch cards and batch processing.
> While many programmers embraced the interactivity and immediate feedback of terminals, Dijkstra argued that the "trial and error" approach fostered by interactive systems led to sloppy thinking and poor program design. He believed that the batch processing environment, which necessitated careful, error-free coding before submission, instilled the discipline necessary for writing robust, well-thought-out code.
> "On the Cruelty of Really Teaching Computing Science" (EWD 1036) (1988 lecture/essay)
Seriously, the laments I hear now have been the same in my entire career as a computer scientist. Let's just look toward to 2035 where someone on HN will complain some old way of doing things is better than the new way because its harder and wearing hair shirts is good for building character.
For an experienced engineer, working out the syntax, APIs, type issues, understanding errors, etc is the easy part of the job. Larger picture issues are the real task.
But for many Jr engineers it’s the hard part. They are not (yet) expected to be responsible for the larger issues.
what is a larger issue? lacking domain knowledge? or lacking deeper understanding of years of shit in the codebase that seniors may have better understanding? where I work, there is no issue that it "too large" for a junior to take on, it is the only way that "junior" becomes "non-junior" - by doing, not by delegating to so-called seniors (I am one of them)
"Larger issue" is overall technical direction and architecture, making decisions that don't paint you into a corner, establishing maintainability as a practice, designing work around an organization's structure and habit and so on.
But these are the things people learn through experience and exposure, and I still think AI can help by at least condensing the numerous books out there around technology leadership into some useful summaries.
You can't give a junior tasks that require experience and nuance that have been acquired over years of development. If you babysit them, then perhaps but then what is the point? By it's nature "nuance" is something hard to describe concretely but as someone who has mentored a fair few juniors most of them don't have it. AI generally doesn't have it either. Juniors need tasks at the boundary of their capability, but not far beyond to be able to progress. Simply allowing them to make a mess of a difficult project is not a good way to get there.
There is such a thing as software engineering skill and it is not domain knowledge, nor knowledge of a specific codebase. It is good taste, an abstract ability to create/identify good solutions to a difficult problem.
Unnecessary complexity, completely arbitrary one off designs, over emphasis on one part of the behavior while ignoring others. Using design patterns where they shouldn't be used, code once and forget operations exist, using languages and framework that are familiar but unfit for that purpose. The list goes on and I see it happen all the time, AI only makes it worse because it tend to verify all of these with "You're absolutely correct!".
And this is always my question: "... because the genie, used well, accelerates learning." Does it though?
How are we defining "learning" here? The example I like to use is that a student who "learns" what a square root is, can calculate the square root of a number on a simple 4 function calculator (x, ÷, +, -) if iteratively. Whereas the student who "learns" that the √ key gives them the square root, is "stuck" when presented with a 4 function calculator. So did they 'learn' faster when the "genie" surfaced a key that gave them the answer? Or did they just become more dependent on the "genie" to do the work required of them?
You still need to be curious. I learn a ton by asking questions of the LLMs when I see new things. “Explain this to me - I get X but why did you do Y?”
It’s diamond age and a half - you just need to continue to be curious and perhaps slow your shipping speed sometimes to make sure you budget time for learning as well.
We had 3 interns this past summer - with AI I would say they were VERY capable of generating results quickly. Some of the code and assumptions were not great, but it did help us push out some releases quickly to alleviate customer issues. So there is a tradeoff with juniors. May help quickly get features out, may also need some refactoring later.
I think the big win with AI is being able to work around jargon. Don't know what that word means ask AI. what the history on it no problem. don't understand a concepts explain this at a high school reading level.
I hate to be so negative, but one of the biggest problems junior engineers face is that they don't know how to make sense of or prioritize the gluttony of new-to-them information to make decisions. It's not helpful to have an AI reduce the search space because they still can't narrow down the last step effectively (or possibly independently).
There are junior engineers who seem to inherently have this skill. They might still be poor in finding all necessary information, but when they do, they can make the final, critical decision. Now, with AI, they've largely eliminated the search problem so they can focus more on the decision making.
The problem is it's extremely hard to identify who is what type. It's also something that senior level devs have generally figured out.
The cynic in me sees it as using juniors as a vehicle for driving up AI metrics. The seniors will be less critical reviewing AI output with a human shield/messenger.
I kind of agree here. The mental model that works for me is "search results passed through a rock tumbler". Search results without attribution and mixed-and-matched across reputable and non-reputable sources, with a bias toward whatever source type is more common.
Which was trained on a pre-AI internet. What's going to happen in coming years when new tech comes out but perhaps isn't documented the same way anymore? It's not an unsolvable problem, but we could see unintended consequences like, say where you must pay the AI provider to ingest your data. Similar to buying poll space or AdSense or whatever they call it for search engines
If you release a new piece of technology from 2025 onwards and don't invest a decent amount of effort into producing LLM-friendly documentation (with good examples) that a user can slurp into their coding agent you're doing your new technology a disservice.
I've learned a lot of shit while getting AI to give me the answers, because I wanted to understand why it did what it did. It saves me a lot of time trying to fix things that would have never worked, so I can just spend time analyzing success.
There might be value in learning from failure, but my guess is that there's more value in learning from success, and if the LLM doesn't need me to succeed my time is better spent pushing into territory where it fails so I can add real value.
>I've learned a lot of shit while getting AI to give me the answers
I would argue you're learning less than you might believe. Similarly to how people don't learn math by watching others solve problems, you're not going to learn to become a better engineer/problem solver by reading the output of ChatGPT.
What you usually do is follow the book instructions and get some result, then go to do some exploration on your own. There’s no walk in the dark trying to figure your own path.
Once you learn what works, and what does not, then you’ll have a solid foundation to tackle more complex subject. That’s the benefit of having a good book and/or a good teacher to guide you to the path of mastering. Using a slot machine is more tortuous than that.
I don't find it to be more torturous than that. In fact, if I were to go back and learn lisp again, I think I'd be a lot more motivated seeing how to build something interesting out of the gate rather than the toy programs I learned in my racket course.
Also, for a lot of things, that is how people learn because there aren't good textbooks available.
I was helping a few people on getting started with an Android Development bootcamp and just being able to run the default example and get their bearing around the IDE was interesting to them. And I remember when I was first learning python. Just doing basic variable declaration and arithmetic was interesting. Same with learning C and being able to write tic-tac-toe.
I think a lot of harm is being done by making beginner have expectations that would befit people that have years of experience. Like you can learn docker in 2 months to someone that doesn't even know Linux exists or have never encountered the word POSIX.
Don’t confuse this with this persons ability to hide their instincts. He is redefining “senior” roles as junior, but words are meaningless in a world of numbers. The $$$ translation is that something that was worth $2 should now be worth $1.
I disagree. In my experience AI does most of the work and the juniors already poor skills atrophy. Then a senior engineer has to review AI slop and tell the junior to roll the AI dice again.
first response from me "let me mention how the real business world actually works" .. let's add a more nuanced slice to that however
Since desktop computers became popular, there have been thousands of small to mid-size companies that could benefit from software systems.. A thousand thousand "consultants" marched off to their nearest accountant, retailer, small manufacturer or attorney office, to show off the new desktop software and claim ability to make new, custom solutions.
We know now, this did not work out for a lot of small to mid-size business and/or consultants. Few could build a custom database application that is "good enough" .. not for lack of trying.. but pace of platforms, competitive features, stupid attention getting features.. all of that, outpaced small consultants .. the result is giant consolidation of basic Office software, not thousands of small systems custom built for small companies.
What now, in 2025? "junior" devs do what? design and build? no. Cookie-cutter procedures at AWS lock-in services far, far outpace small and interesting designs of software.. Automation of AWS actions is going to be very much in demand.. is that a "junior dev" ? or what?
This is a niche insight and not claiming to be the whole story.. but.. ps- insert your own story with "phones" instead of desktop software for another angle
One thing I'd point out is that there are only so many ways to write a document or build a spreadsheet. There are a ton of business processes that are custom enough to that org that they have to decide to go custom, change their process, or deal with the inefficiency of not having a technical solution that accomplishes the goal easily.
Lotus Notes is an example of that custom software niche that took off and spawned a successful consulting ecosystem around it too.
certainly no -- not "all software" of anything. Where is the word "enterprise" in the post you have replied to ? "enterprise" means the very largest companies and institutions..
I did not write "all software" or "enterprise software" but you are surprised I said that... hmmm
I'm not swayed by appeals to authority, but this is a supremely bad take.
"AI" tools are most useful in the hands of experienced developers, not juniors. It's seniors who have the knowledge and capability to review the generated output, and decide whether the code is bullshit and will only cause more issues when it's merged, or whether it's usable if they tweak and adapt it in certain ways.
A junior developer has no such skills. Their only approach will be to run the code, test whether it fulfills the requirements, and, if they're thorough, try to understand and test it to the best of their abilities. Chances are that because they're pressured to deliver as quickly as possible to impress their colleagues and managers, they'll just accept whatever working solution the tool produces the first time.
This makes "AI" in the hands of junior developers extremely dangerous. Companies who allow this type of development will quickly grind to a halt under the weight of technical debt, and a minefield of bugs they won't know how to maneuver around.
The sad reality is that with "AI" there is no pathway for junior developers to become senior. Most people will gravitate towards using these tools as a crutch for quickly generating software, and not as a learning tool to improve their own skills. This should concern everyone vested in the future of this industry.
The dude literally invented Extreme Programming and was the first signer of the Agile Manifesto. He's forgotten more about software development than most people on this site ever knew.
Someone's accomplishments don't make them incapable of having bad opinions and being wrong. Cults of personality are harmful to progress. Opinions should hold the same weight and be held to the same scrutiny regardless of who voiced them.
That wasn't the question being asked. The question being asked was literally "what are this guy's accomplishments," and Kent Beck is a tech industry OG with a laundry list of them.
Of course he can be wrong; he's human. That wasn't my point.
To be fair, even if I appreciate Beck, some people do get too famous and start to inhabit a space that is far removed from the average company. Many of these guys tend to give out advice that is applicable to a handful of top earning companies but not the rest.
Interesting take... I'm seeing a pattern... People think AI can do it all... BUT I see juniors often are the ones who actually understand AI tools better than seniors... That's what AWS CEO points out... He said juniors are usually the most experienced with AI tools, so cutting them makes no sense... He also mentioned they are usually the least expensive, so there's little cost saving... AND he warned that without a talent pipeline you break the future of your org... As someone who mentors juniors, I've seen them use AI to accelerate their learning... They ask the right questions, iterate quickly, and share what they find with the rest of us... Seniors rely on old workflows and sometimes struggle to adopt new tools... HOWEVER the AI isn't writing your culture or understanding your product context... You still need people who grow into that... So I'm not worried about AI replacing juniors... I'm more worried about companies killing their own future talent pipeline... Let the genies help, but don't throw away your apprentices.
ON TOP OF IT ALL, juniors are the ones who bring novel tools to the desk MOST times...i.e. I had no clue the Google IDE gave you free unlimited credits despite the terrible UI...but a young engineer told me about it!!
I've seen seniors and juniors bring novel tools in. Seniors do it less often perhaps - but only because we have seen this before under a different name and realize it isn't novel. (sometimes the time has finally come, sometimes it fail again for the same reason it failed last time)
I've seen seniors bring up novel ideas to juniors—well, novel to the juniors anyway.
Just an example. I've been in so many code bases over the years… I had a newer engineer come aboard who, when he saw some code I recently wrote with labels (!) he kind of blanched. He thought "goto == BAD". (We're talking "C" here.)
But this was code that dealt with Apple's CoreFoundation. More or less every call to CF can fail (which means returning NULL in the CF API). And (relevant) passing NULL to a CF call, like when trying to append a CF object to a CF array, was a hard crash. CF does no param checking. (Why, that would slow it down—you, dear reader, are to do all the sanity checking.)
So you might have code similar to:
CFDictionary dict = NULL;
dict = CFCreateDictionary();
if (!dict)
goto bail;
You would likely continue to create arrays, etc—insert them into your dictionary, maybe return the dictionary at the end. And again, you checked for NULL at every call to CF, goto bail if needed.
Down past 'bail' you could CFRelease() all the non-null instances that you do not return. This was how we collected our own garbage. :-)
In any event, goto labels made the code cleaner: your NULL-checking if-statements did not have to nest crazy deep.
The new engineer admitted surprise that there might be a place for labels. (Or, you know, CF could have been more NULL-tolerant and simply exited gracefully.)
I'm just shocked people aren't clueing into the fact that tech companies are trying to build developer dependence on these things to secure a "rent" revenue stream. But hey, what do I know. It's just cloud hyper scaling all over again. Don't buy and drive your own hardware. Rent ours! Look, we built the metering and everything!
I'd hope people are. It's painfully obvious this entire AI push is rent-seeking half hidden by big tech money. At some point the free money is going to go away, but the rent for every service will remain.
Their implication is that junior devs have more likely built up their workflow around the use of AI tooling, likely because if they're younger they'll have had more plasticity in their process to adapt AI tooling
Overall I don't quite agree. Personally this applies to me, I've been using vim for the last decade so any AI tooling that wants me to run some electron app is a non starter. But many of my senior peers coming from VS Code have no such barriers
Speaking of vim - adding and configuring copilot plugin for vim is easy (it runs a nodejs app in the background but if you have spare 500 Mb RAM it's invisible).
Speaking as a senior dev, anecdotally juniors may indeed understand AI tools better, because they spend more hours a day coding and working with the tools, and they need the tools to understand the codebase or to be productive. Seniors have more hours stuck in meetings, developing specs/tickets for the juniors, code reviewing, etc. Seniors are likely to not bother with a prompt for simple changes in codebases they already understand.
Amazon has an internal platform for building software. The workflows are documented and have checks and balances. So the CEO wants to have more junior developer that are more proficient with AI, and have (in ratio) less senior developers. Also, product context comes from (product) managers, UX designers.
For medium or small companies, these guardrails or documentation can be missing. In that case you need experienced people to help out.
Nah, models can be fine tuned and trained on anything. Common consumer products like ChatGPT and Gemini have particular styles, very polite and helpful, but there are models trained to be combatative, models trained to write in the style of shakespeare, all sorts of things. Someone could train a model to reply to posts in the style of HN comments and you’d probably never know.
7/10 senior devs (usually fellas 50+) will get mad at you for trying to use Claude Code. Me: “dude it writes better code than crap you write in your mush middle-age brain.” Also me: “I also have mush brain.”
I think LLM is a reflection of human intelligence. If we humans become dumber as a result of LLM, LLMs will also become dumber. I’d like to think in some dystopian world, LLM’s trained from pre 2023 data will be sought after.
you're right but my opinion about this has changed
I would have agreed with you 100% one year ago. Basically senior engineers are too complacent to look at AI tools as well as ego driven about it, all while corporate policy disincentivizes them from using anything at all, with maybe a forced Co-Pilot subscription. While junior engineers will take a risk that the corporate monitoring of cloud AI tools isn't that robust.
But now, although many of those organizations are still the same - with more contrived Co-Pilot subscriptions - I think senior engineers are skirting corporate policy too and become more familiar with tools.
I'm also currently in an organization that is a total free for all with as many AI coding and usage tools as necessary to deliver faster. So I could be out of touch already.
Perhaps more complacent firms are the same as they were a year ago.
Maybe, but you make it sound like juniors are more worthy to companies than seniors. Then fire most/all seniors and good luck with resulting situation.
Coding in any sufficiently large organization is never the main part of senior's time spend, unless its some code sweatshop. Juniors can do little to no of all that remaining glue that makes projects go from a quick brainstorming meeting to live well functioning and supported product.
So as for worth - companies can, in non-idedal fashion obviously, work without juniors. I can't imagine them working without seniors, unless its some sort of quick churn of CRUDs or eshops from some templates.
Also there is this little topic that resonates recently across various research - knowledge gained fast via llms is a shallow one, doesn't last that long and doesn't go deeper. One example out of many - any time I had to do some more sophisticated regex-based processing I dived deep into specs, implementation etc. and few times pushed it to the limits (or realized task is beyond what regex can do), instead of just given the result, copypasted it and moved along since some basic test succeeded. Spread this approach across many other complex topics. That's also a view on long term future of companies.
I get what you say and I agree partially but its a double edged sword.
I think the temptation to use AI is so strong that it will be those who will keep learning who will be valuable in future. Maybe by asking AI to explain/teach instead of asking for solution direclty. Or not using AI at all.
> I'm pretty sure AI is going to lead us to a deskilling crash.
That's my thought too. It's going to be a triple whammy
1. Most developers (Junior and Senior) will be drawn in by the temptation of "let the AI do the work", leading to less experience in the workforce in the long term.
2. Students will be tempted to use AI to do their homework, resulting in new grads who don't know anything. I have observed this happen first hand.
3. AI-generated (slop) code will start to pollute Github and other sources used for future LLM training, resulting in a quality collapse.
I'm hoping that we can avoid the collapse somehow, but I don't see a way to stop it.
I think seniors know enough to tell whether they need to learn or not. At least that's what I tell myself!
The thing with juniors is: those who are interested in how stuff works now have tools to help them learn in ways we never did.
And then it's the same as before: some hires will care and improve, others won't. I'm sure that many juniors will be happy to just churn out slop, but the stars will be motivated on their own to build deeper understanding.
On the contrary, being able to access (largely/verifiably) correct solutions to tangible & relevant problems is an extremely great way to learn by example.
It should probably be supplemented with some good old RTFM, but it does get us somewhat beyond the "blind leading the blind" StackOverflow paradigm of most software engineering.
I can't help but feel this is backpedaling after the AI hype led to people entering university avoiding computer science or those already in changing their major. Ultimately we might end up with a shortage of developers again, which would be amusing.
I went to university 2005-2008 and I was advised by many people at the time to not go into computer science. The reasoning was that outsourced software developers in low-cost regions like India and SEA would destroy salaries, and software developers should not expect to make more than $50k/year due to the competition.
Even more recently we had this with radiologists, a profession that was supposed to be crushed by deep learning and neural networks. A quick Google search says an average radiologist in the US currently makes between $340,000 to $500,000 per year.
This might be the professional/career version of "buy when there's blood in the streets."
You nailed it on the head, down to the exact examples.
I was still in high school in 2010, and was told the same thing about outsourcing to India/SEA/etc. making a CS degree/career (in the US) a terrible choice. It wasn't just random people saying this either, I was reading about it in the news, online, had some family acquaintances with alleged former software dev career, etc. I didn't listen, and I am glad I didn't.
As I was graduating from college, and deep learning was becoming a new hot thing, I heard the same thing about radiologists, and how they are all getting automated away in the next 5 years. I had no plans to go to med school, and I didn't know anyone at the time who went through it, so I didn't know much about the topic. On the surface, it seemed like a legitimate take, and I just stored it in my head as "sounds about right."
Cue to now, I know more than a few people who went through med school, and am in general more attuned to the market. Turns out, all of that was just another genpop hype, those news articles about "omg radiologists are all getting replaced by computers" stopped from showing up on any of my news feeds, and not a single radiology-specialized med school graduate I know had any issues with getting a job (that paid significantly more than an entry level position at a FAANG).
I have zero idea what point I was trying to make with this comment, but your examples mirror my personal experience with the topic really well.
I went for CS in my late 20s, always tinkered with computers but didn't get into programming earlier. College advisor told me the same thing, and that he went for CS and it was worthless. This was 2012.
I had a job lined up before graduating. Now make high salary for the area, work remotely 98% of the time and have flexible schedule. I'm so glad I didn't listen to that guy.
The one thing I learned in college is that the advisors are worthless. There's how many students? And you are supposed to expect they know the best thing for you? My advisor told me that all incoming freshmen must take a specific math class, a pre-calculus course, totally ignoring all of my AP exams that showed I was well beyond that. Wasted my time and money.
My take is that these are not binary issues. With outsourcing, it is true that you can hire someone cheaper in Asian countries but it cannot kill all jobs locally. So what happens is that the absolute average/mediocre get replaced by outsourcing and now with AI while the top talent can still command a good salary because they are worth it.
So I think that a lot of juniors WILL get replaced by AI not because they are junior necessarily but because a lot of them won't be able to add great value compared to a default AI and companies care about getting the best value from their workers. A junior who understands this and does more than the bare minimum will stand out while the rest will get replaced.
> So I think that a lot of juniors WILL get replaced by AI not because they are junior necessarily but because a lot of them won't be able to add great value compared to a default AI and companies care about getting the best value from their workers. A junior who understands this and does more than the bare minimum will stand out while the rest will get replaced.
Again this is what people said about outsourced developers. 2008 logic was, why would anyone hire a junior for $50k/year when you could hire a senior with 20 years experience for $10k/year from India?
Reality: for 5+ years you could change careers by taking a 3-6 month JavaScript bootcamp and coming out the other end with a $150k job lined up. That's just how in demand software development was.
> Even more recently we had this with radiologists, a profession that was supposed to be crushed by deep learning and neural networks. A quick Google search says an average radiologist in the US currently makes between $340,000 to $500,000 per year.
At the end of the day, radiologists are still doctors.
Yup hearing big talk about competition and doom is a strong signal that there is plenty of demand.
You can either bet on the new unproven thing claiming to change things overnight, or just do the existing thing that's working right now. Even if the new thing succeeds, an overnight success is even more unrealistic. The insight you gain in the meantime is valuable for you to take advantage of what that change brings. You win either way.
When there is no competition that is a sign there is no demand.
There can sometimes be too much competition, but often there is only the illusion of too much if you don't look at quality. You can find a lot of cheap engineers in India, but if you want a good quality product you will have to pay a lot more.
Can anyone really blame the students? If I were in their shoes, I probably wouldn't bother studying CS right now. From their perspective, it doesn't really matter whether AI is bullshit in any capacity; it matters whether businesses who are buying the AI hype are going to hire you or not.
Hell, I should probably be studying how to be a carpenter given the level at which companies are pushing vibe coding on their engineers.
"after the AI hype led to people entering university avoiding computer science or those already in changing their major"
That's such a terrible trend.
Reminds me of my peers back in ~2001 who opted not to take a computer science degree even though they loved programming because they thought all the software engineering jobs would be outsourced to countries like India and there wouldn't be any career opportunities for them. A very expensive mistake!
Certainly, I even know of experienced devs switching out of tech entirely. I think the next couple of decades are going to be very good for software engineers. There will be an explosion of demand yet a contraction in supply. We are in 2010 again.
There will be programmers of the old ways, but AI is basically code 2.0, there are now a lot of things that are AI specific that those with traditional software development skills can’t do.
It's backpedaling but I don't think it's planning ahead to prevent a developer shortage - rather it's pandering to the market's increasing skepticism around AI and that ultimately the promised moonshot of AI obsoleting all knowledge work didn't actually arrive (at least not in the near future).
It's similar to all those people who were hyping up blockchain/crypto/NFTs/web3 as the future, and now that it all came to pass they adapted to the next grift (currently it's AI). He is now toning down his messaging in preparation of a cooldown of the AI hype to appear rational and relevant to whatever comes next.
You are right, perfect amount of false humility and balance. The wage suppression is an accidental biproduct and not the intent. Collateral damage if you will.
Or maybe they realize the AI needs humans in the loop for the foreseeable future for enterprise use cases and juniors (and people from LCL areas) are cheaper and make the economics make some sort of sense.
So he's saying we should be replacing the seniors with fresh grads who are good at using AI tools? Not a surprising take, given Amazon's turnover rate.
My experience is that juniors have an easier time to ramp up, but never get better at proper engineering (analysis) and development processes (debug). They also struggle to read and review code.
I fear that unless you heavily invest in them and follow them, they might be condemned to have decades of junior experience.
In my view there's two parts to learning, creation and taste, and both need to be balanced to make progress. Creation is, in essence, the process of forming pathways that enable you to do things, developing taste is the process of pruning and refining pathways to doing things better.
You can't become a chef without cooking, and you can't become a great one without cultivating a taste (pun intended) for what works and what it means for something to be good.
From interactions with our interns and new-grads, they lack the taste, and rely too much on the AI for generation. The consequence is that when you have conversations with them, they straggle to understand the concepts and tools they are using because they lack the familiarity that comes with creation, and they lack the skills to refine the produced code into something good.
A Junior Dev is not just for Christmas, it’s for life.
But more seriously are there CEOs out there who think they can replace the people starting off in their industry with AI? Who do the think will be the senior devs in 5-10yrs?
We frequently get juniors or interns who are perfectly capable of pumping out many LoC with the use of AI in various forms - the issue is that they _don't_ actually ever learn how to think for themselves, and can't fix problems when something goes wrong or the LLM paints itself into a corner. I have found myself doing a lot more shepherding and pairing with juniors when they can't figure something out recently, because they just have not had the space to build their own skills.
1. replacing junior engineers, with AI ofcourse breaks the talent pipeline. Seniors will retire one day, who is going to replace them? Are we taking the bet, that we wont need any engineer at that time? sounds dangerous.
2. Junior engineer's heavy reliance on AI tools is a problem in itself. AI tools learn from existing code that is written by senior engineers. Too much use of AI by junior engineers will result in deterioration of engineering skills. It will eventually result in AI learning from AI generated code. This is true for most other content as well, as more and more content on internet is AI generated.
I believe the idea is to not stop hiring juniors. Instead it's to replace anybody that commands a high salary with a team of cheaper juniors armed with LLM's. The idea is more about dragging down average pay than never hiring anybody. At least for now.
And then all those unemployed seniors with extensive domain knowledge use AI to speedrun the creation of competition and you need to spend $$$$ to buy them out and shut them down. Solid idea.
I have heard this thing quite a few times over last few months each time is Amazon or AWS CEOs. May be this time he want to replace senior engineers. That would be more useful for them as each passing year they more and more of them and in times like these they are not looking to go leave Amazon on their own.
These people are working on destroying the planet to make more money, they absolutely do not care. Our society isn't set up to punish them, but encourage such behavior to even more extremes (see datacenter build outs causing water shortages, electricity hikes, and cancer in poor communities; nearly every politician capitulating on such actions because they don't know better).
I wish people would get off the "AI is the worst thing for the environment" bandwagon. AI and data centers as a whole aren't even in the top 100 emitters of pollution and never will be.
If you want to complain about tech companies ruining the environment, look towards policies that force people to come into the office. Pointless commutes are far, far worse for the environment than all data centers combined.
Complaining about the environmental impact of AI is like plastic manufacturers putting recycling labels on plastic that is inherently not recycleable and making it seem like plastic pollution is every day people's fault for not recycling enough.
AI's impact on the environment is so tiny it's comparable to a rounding error when held up against the output of say, global shipping or air travel.
Why don't people get this upset at airport expansions? They're vastly worse.
Of course they aren't polluters as in generating some kind of smoke themselves. But they do consume megawatts upon megawatts of power that has to be generated somewhere. Not often you have the luxury of building near nuclear power plant. And in the end you're still releasing those megawatts as heat into the atmosphere.
Obviously after 10-15 years of experience working as a developer AI will be a senior dev. Probably will get promoted to management with all that experience.
Promoting your best engineers to management sometimes gets you a great manager, but often gets you a mediocre or just-about-competent manager at the cost of a great engineer.
I'm a big fan of the "staff engineer" track as a way to avoid this problem. Your 10-15 year engineers who don't vibe with management should be able to continue earning managerial salaries and having the biggest impact possible.
I'm also a fan of leadership without management. Those experienced engineers should absolutely be taking on leadership responsibilities - helping guide the organization, helping coach others, helping build better processes. But they shouldn't be stuck in management tasks like running 1-1s and looking after direct reports and spending a month every year on the annual review process.
This is a general problem that corporations have trouble with with: The struggle to separate leadership and people management. Why does the person who tells you what to do also need to be the same person who does your annual review, who also has to be the same person who leads the technical design of the project, approves your vacation, assists with your career development, and gives feedback or disciplinary correction when you mess up? Why do we always seem to bundle all these distinct roles together under "Manager"?
This is exactly where I find myself. I've been asked several times to take on management, but I have no interest in it. I got to be a principal after 18 years of experience by being good at engineering, not management. Like you said, I can and do help with leadership through mentorship, offering guidance and advice, giving presentations on technical topics, and leading technical projects.
Absolutely agree. Regardless, my org keeps trying to get me to take a management role after 15 years dev experience. I love my job and don't like managing people. You couldn't pay me enough to become a manager.
I gave opus an "incorrect" research task (using this slash command[1]) in my REST server to research to use SQLite + Litestream VFS can be used to create read-replicas for REST service itself. This is obviously a dangerous use of VFS[2] and a system like sqlite in general(stale reads and isolation wise speaking). Ofc it happily went ahead and used Django's DB router feature to implement `allow_relation` to return true if `obj._state.db` was a `replica` or `default` master db.
Now claude had access to this[2] link and it got the daya in the research prompt using web-searcher. But that's not the point. Any Junior worth their salt — distributed systems 101 — would know _what_ was obvious, failure to pay attention to the _right_ thing. While there are ideas on prompt optimization out there [3][4], the issue is how many tokens can it burn to think about these things and come up with optimal prompt and corrections to it is a very hard problem to solve.
Ive been managing and supporting teams for a long time and i'm sorry, but junior and mid-level devs do the majority of the heavy lifting when it comes to work output in big corps. I don't think AI will replace them. I don't think all these IC5 and IC6 engineers are going to be putting up 400-500 diffs a year anytime soon.
I recently pair-worked with two junior developers (on their first job, but still with like 2+ years with the company) in order to transfer the know-how of something.
I realized that they are shockingly bad at most basic things. Still their PR:s look really good (on the surface). I assume they use AI to write most of the code.
What they do excel in is a) cultural fit for the company and b) providing long-term context to the AIs for what needs to be done. They are essentially human filters between product/customers and the AI. They QA the AI models' output (to some extent).
But I think the actual reason was not addressed. The work of junior devs is exactly what can be replaced by AI, instead of the more complex abilities senior development possess.
The level of cynicism here is astronomical. After discovering the strategy of "fire juniors and let a few seniors manage autonomous agents" was an abject failure, now the line is "actually juniors are great because we've brainwashed them into thinking AI is cool and we don't have to pay them so much". Which makes me want to vomit.
The only relevant point here is keeping a talent pipeline going, because well duh. That it even needs to be said like it's some sort of clever revelation is just another indication of the level of stupid our industry is grappling with.
Now with AI, I expect junior developers to learn much quicker and progress to senior very quickly. I'd now rather hire at least 1 of each to begin with, both "junior" and a "senior" developer and then additionally hire more juniors to quickly turn them into a "senior".
We do not need to hire anymore outside senior developers who need to be trained on the codebase with AI, given that the junior developers catch up so quickly they already replaced the need to hire a senior developer.
Therefore replacing them with AI agents was quite premature if not completely silly. In fact it makes more sense to hire far less senior developers and to instead turn juniors directly into senior developers to save lots of money and time to onboard.
To me the more insidious problem is that we have juniors now that aren’t learning much because they lean on AI for everything. They are behind the curve.
This is performative bullshit pandering to the increased skepticism around AI. He wouldn't be saying that if AI investment was still in full swing.
I do agree with him about AI being a boon to juniors and pragmatic usage of AI is an improvement in productivity, but that's not news, it's been obvious since the very beginnings of LLMs.
So it's performative when the head of AWS says it and not news. But it's not performative when you say it and people should have listened to you in the comments?
It's performative when you talk whatever the market wants to hear rather than sticking to an opinion (no matter how flawed it is). This behavior reminds me of the cryptobros that were hailing NFTs/web3 as the next best thing since sliced bread, and when that didn't came to pass quietly moved onto the next grift (AI) with the same playbook.
(also I’m just talking out of my ass on a tech forum under a pseudonym instead of going to well-publicized interviews)
They must’ve forgot who created the first tech hype bubbles in the first place bc I’m about to replace some of these companies if I don’t get hired soon
Most of the apps that I use regularly fail at least once a day nowadays. I think this is a direct cause of putting AI code in production without reviewing/QA.
While I have no particular love for AI generated code, I think this has nothing to do with AI. Software has been unreliable for over a decade. Companies have been rushing out half baked products and performing continual patches for many years. And it's our fault because we have just come to accept it.
The "over" deserves a lot of emphasis. To this day, I save my code at least once per line that I type because of the daily (sometimes hourly) full machine crashes I experienced in the 80s and 90s.
The problem is human, not technical. Companies and managers need to start caring about the details instead of crossing items off a list. Until we see that culture shift in the industry, which might never happen, AI isn't going to help—if anything, it'll make the problem worse as devs rush to deliver on arbitrary deadlines.
Well the reason I think it might be different is that I am noticing a material change in my behavior.
I have always cared a lot about quality and craftsmanship. Now when I am working and notice something wrong, I just fix it. I can code it entirely with AI in the time it would've take me to put it on an eternal backlog somewhere.
Plus, if you are skipping tests or telling yourself, you wrote them when they don’t actually verify anything in the first place, then buying into a hype cycle of “the AI writes perfect code“ is unlikely to break the pattern
This sounds like a comment from someone who doesn't have visibility into how good the models are getting and how close they are to fully autonomous, production-grade software development.
This is an easy theory to prove; if AI was anywhere close to a senior engineer, we'd see the costs of software development drop by a corresponding amount or quality would be going up. Not to mention delivery would become faster. With LLMs being accessible to the general public I'd also expect to see this in the open-source world.
I see none of that happening - software quality is actually in freefall (but AI is not to blame here, this began even before the LLM era), delivery doesn't seem to be any faster (not a surprise - writing code has basically never been the bottleneck and the push to shove AI everywhere probably slows down delivery across the board) nor cheaper (all the money spent on misguided AI initiatives actually costs more).
It is a super easy bet to take with money - software development is still a big industry and if you legitimately believe AI will do 90% of a senior engineer you can start a consultancy, undercut everyone else and pocket the difference. I haven’t heard of any long-term success stories with this approach so far.
This sounds like a comment from someone who has tested it in a limited capacity such as small blog sites or side projects that did not need to be maintained
The thing people miss in these “replace juniors with AI” takes is that juniors were never mainly about cheap hands on keyboards. They’re the only people in the org who are still allowed to ask “dumb” questions without losing face, and those questions are often the only signal you get that your abstractions are nonsense.
What AI does is remove a bunch of the humiliating, boring parts of being junior: hunting for the right API by cargo-culting Stack Overflow, grinding through boilerplate, getting stuck for hours on a missing import. If a half-decent model can collapse that search space for them, you get to spend more of their ramp time on “here’s how our system actually fits together” instead of “here’s how for-loops work in our house style”.
If you take that setup and then decide “cool, now we don’t need juniors at all”, you’re basically saying you want a company with no memory and no farm system – just an ever-shrinking ring of seniors arguing about strategy while no one actually grows into them.
Always love to include a good AI x work thread in my https://hackernewsai.com/ newsletter.
"without losing face"? What culture are you referring to? The Western companies I have worked at do not discourage such questions -- in fact, it's often the sign of someone very senior when they ask a seemingly 'dumb' question that others have taken for granted.
Completely agree with this. I got to work closely with an IBM fellow one summer and I was impressed by his willingness to ask "dumb questions" in meetings. Sometimes he was just out of the loop but more often he was just questioning some of the assumptions that others in the room had left unquestioned.
Unfortunately, I found that the culture of "think." at IBM is not matched at many other organizations. Most days, I miss it.
But forced RTO and only 10 days off per year is enough to keep me away ;)
Company culture. Some companies I worked for would fire you for questioning decisions. Others, welcomed criticism. You don’t really know which environment you’re in until someone says something. Are you going to take the risk and be the first?
Yes, because i would rather not work at such company and go somewhere else.
This is a really good and under appreciated point. My recommendation to mid-level, senior, and staff engineers is to keep questioning decisions and create a culture where that’s encouraged.
Junior devs do that naturally (if you have the culture) because they don’t know anything already. It’s great
> My recommendation to mid-level, senior, and staff engineers is to keep questioning decisions and create a culture where that’s encouraged.
Tell me you've never worked at FAANG without telling me you've never worked at FAANG...
Some of us intentionally avoid FAANG for that reason.
Usually the people who question decisions are shot down because they don’t have a wholistic understanding of the decision and (respectfully) don’t have good arguments. This is only because they are focused on some narrow aspect of the business which distorts or reduces their visibility and understanding.
> This is only because they are focused on some narrow aspect of the business
Is this a bad thing though? If some technical decision has downside risk, I’d reasonably expect:
- the affected stakeholder to bring it up
- the decision maker to assuage the stakeholder’s concern (happy path) or triage and escalate
I think you are right. It's still worth encouraging people to question decisions even though most of the time it won't be the right compromise for the business.
One of the most regulated industries, aviation requires crews to go through crew management training where it's explicitly trained for lower level staff to raise concerns in spite of perceived superior knowledge.
Some of the biggest accidents have happened directly due to this. Like Tenerife where the flight engineer had been listening to the radio and raised doubts about the runway being free but was ignored by the overconfident captain.
I don't know how to calculate "usually", in my experience people who question decisions in my company are shot down because the decision makers are usually (no pun intended) very incompetent and the questions make that visible, even if not intended. Many companies that I know are so corrupt that competent people are considered to be dangerous for the status quo.
> They’re the only people in the org who are still allowed to ask “dumb” questions without losing face
I strongly disagree, a Senior who cannot ask a "dumb question" is a useless developer to me. Ask away, we want to figure things out, we're not mind readers. Every good senior developer I know asks questions and admits when they don't know things. This is humility and in my eyes is a strong market for a good Senior Developer. The entire point of our job is to ask questions (to ourselves typically the most) and figure out the answers (the code).
Juniors could or should ask more questions, and by the time you're a Senior, you're asking key questions no matter how dumb they sound.
I think the biggest challenge now becomes how more seasoned engineers teach juniors. The AI makes the ramp a lot easier but you still do best when you understand the whole stack and make mistakes.
It’s damned near impossible to figure out where to spend your time wisely to correct an assumption a human made vs. an AI on a blended pull request. All of the learning that happens during PR review is at risk in this way and I’m not sure where we will get it back yet. (Outside of an AI telling you - which, to be fair, there are some good review bots out there)
Junior engineers now learn from AIs. And AIs now learn from RL cost functions. And RL cost functions are being set by PhDs, with little to no production grade engineering experience ;)
The result is interesting. First, juniors are miserable. What used to be a good experience coding and debugging, in a state of flow is now anxiously waiting if an AI could do it or not.
And senior devs are also miserable, getting apprentices used to be fun and profit, working with someone young is uplifting, and now it is gone.
The code quality is going down, Zen cycle interrupted, with the RL cost functions now at the top.
The only ones who are happy are hapless PhDs ;$
Agree.
> cargo-culting Stack Overflow
What do you mean by this? I understand “cargo-culting” as building false idols, e.g. wooden headphones and runways to attract airplanes that never come.
It means to copy code or instructions from a site into your own project without having any comprehension of how or why it works.
example: you have a Windows problem. You search and read that "sfc /scannow" seems a popular answer to Windows problems. You run it without ever understanding what sfc does, whether the tool is relevant to your problem, etc. You are cargo culting a solution.
Cargo-cult in relation to programming exists since late 80s.
http://www.catb.org/jargon/html/C/cargo-cult-programming.htm...
https://en.wikipedia.org/wiki/Cargo_cult_programming
So, I think there are two models.
One is a "one junior per team" model. I endorse this for exactly the reasons you speak.
Another, as I recently saw, was a 70/30 model of juniors to seniors. You make your seniors as task delegators and put all implementation on the junior developers. This puts an "up or out" pressure and gives very little mentorship opportunities. if 70% of your engineers are under 4 years of experience, it can be a rough go.
That second model is basically the hospital model.
You have 1 veteran doctor overseeing 4 learning doctors. For example operating rooms do this, where they will have 4 operating rooms with 4 less experienced anesthesist and then 1 very experienced anesthesist who will rotate between the 4 and is on call for when shit hits the fan.
Honestly I think everyone here is missing the forest for the trees. Juniors their main purpose isn't to "ask questions", it's to turn into capable seniors.
That's also why the whole "slash our junior headcount by 3/4th" we are seeing across the industry is going to massively, massively backfire. AI / LLMs are going to hit a wall (well, they already hit it a while ago), suddenly every is scrambling for seniors but there are none because no one wanted to bear the 'burden' of training juniors to be seniors. You thought dev salaries are insane now? Wait until 4-5 years from now.
With the subjective view on what a junior is, I think the 70-30 - or higher - model is used in any company I ever interacted with. For this evaluation I consider junior=someone with less skills than needed to do the job autonomously/require direction and supervision most time, senior=someone that can work autonomously.
AI will replace jobs. People are putting their IT/dev budget into something, which means something else will be cut.
I also don’t believe juniors, kids, seniors, staff, principals, distinguished/fellow should be replaced by AI. I think they WILL be, but they shouldn’t be. AI at Gemini 3 Flash / Claude Opus 4.5 level is capable with help and review of doing a lot of what a lot of devs do currently. It can’t do everything and will fail, but if the business doesn’t care, they’ll cut jobs.
Don’t waste time trying to argue against AI to attempt to save your job. Just learn AI and do your job until you’re no longer needed to even manage it. Or, if you don’t want to, do something else.
This is actually a super power I have after spending my first part of my career in sales.
I was never formally trained so I just keep asking "why" until someone proves it all the way. Sales itself is also a lot about asking questions that won't come up to find the heart of the thing people actually want... which is just another side of the coin.
I mean, that’s an interesting take, but “having people around to ask dumb questions” is not why most orgs hire juniors.
Yikes. Sounds like you work for a toxic company. The mid and senior level engineers I know all go out of their way to ask the dumb questions. Every junior employee I've mentored, I've told them the main way they can fail is not asking questions early and often. Gotta build a culture that supports questions.
If you are so junior that you are spending hours on a missing import, you may have a claim for disability.
What will eventually pan out is that senior devs will be replaced with junior devs powered by AI assistants. Simply because of the reasons you stated. They will ask the dumb important questions and then after a while, will even solve for them.
Now that their minds are free from routine and boilerplate work, they will start asking more 'whys' which will be very good for the organization overall.
Take any product - nearly 50% of the features are unused and it's a genuine engineering waste to maintain those features. A junior dev spending 3 months on the code base with Claude code will figure out these hidden unwanted features, cull them or ask them to be culled.
It'll take a while to navigate the hierarchy but they'll figure it out. The old guard will have no option but to move up or move out.
How do you suppose the old guard are filling their days now?
At some level, aren’t you describing the age-old process of maturing from junior to mid level to senior in most lines of work, and in most organizations? Isn’t that what advancing in responsibility boils down to: developing subtlety and wisdom and political credibility and organizational context? Learning where the rakes are?
I wish 3 months, or even 3 years, were long enough to fully understand the whys and wherefores and politics of the organizations I cross paths with, and the jungle of systems and code supporting all the kinds of work that happen inside…
Why would Claude code help you find unused features? The end customer uses features, not the AI. I would want to know from the end customer whether a feature is unused, and a Junior with an LLM assistant is not going to be able to tell me that without adding new features to the code base.
Relevant post by Kent Beck from 12th Dec 2025: The Bet On Juniors Just Got Better https://tidyfirst.substack.com/p/the-bet-on-juniors-just-got...
> The juniors working this way compress their ramp dramatically. Tasks that used to take days take hours. Not because the AI does the work, but because the AI collapses the search space. Instead of spending three hours figuring out which API to use, they spend twenty minutes evaluating options the AI surfaced. The time freed this way isn’t invested in another unprofitable feature, though, it’s invested in learning. [...]
> If you’re an engineering manager thinking about hiring: The junior bet has gotten better. Not because juniors have changed, but because the genie, used well, accelerates learning.
Isn't the struggling with docs and learning how and where to find the answers part of the learning process?
I would argue a machine that short circuits the process of getting stuck in obtuse documentation is actually harmful long term...
Isn't the struggle of sifting through a labyrinth of physical books and learning how and where to find the right answers part of the learning process?
I would argue a machine that short-circuits the process of getting stuck in obtuse books is actually harmful long term...
It may well be. Books have tons of useful expository material that you may not find in docs. A library has related books sitting in close proximity to one another. I don't know how many times I've gone to a library looking for one thing but ended up finding something much more interesting. Or to just go to the library with no end goal in mind...
Speaking as a junior, I’m happy to do this on my own (and do!).
Conversations like this are always well intentioned and friction truly is super useful to learning. But the ‘…’ in these conversations seems to always be implicating that we should inject friction.
There’s no need. I have peers who aren’t interested in learning at all. Adding friction to their process doesn’t force them to learn. Meanwhile adding friction to the process of my buddies who are avidly researching just sucks.
If your junior isn’t learning it likely has more to do with them just not being interested (which, hey, I get it) than some flaw in your process.
Start asking prospective hires what their favorite books are. It’s the easiest way to find folks who care.
You're reading a lot into my ellipsis that isn't there. :-)
Please read it as: "who knows what you'll find if you take a stop by the library and just browse!"
I’ll also make the observation that the extra time spent is very valuable if your objective solely is learning, but often the Business™ needs require something working ASAP
This should be possible online, it would be if more journals were open access.
Disagree, actually. Having spent a lot of time publishing papers in those very journals, I can tell you that just browsing a journal is much less conducive to discovering a new area to dive into than going to a library and reading a book. IME, books tend to synthesize and collect important results and present them in an understandable (pedagogical?!) way that most journals do not, especially considering that many papers (nowadays) are written primarily to build people's tenure packets and secure grant funding. Older papers aren't quite so bad this way (say, pre-2000).
You could make much the same observation about online search results.
When I first opened QBasic, <N> years ago, when I was a wee lad, the online QBasic help didn't replace my trusty qbasic book (it supplemented it, maybe), nor did it write the programs for me. It was just there, doing nothing, waiting for me to press F1.
AI, on the other hand...
Well, yes -- this is why I still sit down and read the damn books. The machine is useful to refresh my memory.
learning to learn
I recall similar arguments being made against search engines: People who had built up a library of internal knowledge about where and how to find things didn't like that it had become so easy to search for resources.
The arguments were similar, too: What will you do if Google goes down? What if Google gives the wrong answer? What if you become dependent on Google? Yet I'm willing to bet that everyone reading this uses search engines as a tool to find what they need quickly on a daily basis.
I argue that there is a strong, strong benefit to reading the docs: you often pick up additional context and details that would be missing in a summary.
Microsoft docs are a really good example of this where just looking through the ToC on the left usually exposes me to some capability or feature of the tooling that 1) I was not previously aware of and 2) I was not explicitly searching for.
The point is that the path to a singular answer can often include discovery of unrelated insight along the way. When you only get the answer to what you are asking, you lose that process of organic discovery of the broader surface area of the tooling or platform you are operating in.
I would liken AI search/summaries to visiting only the well-known, touristy spots. Sure, you can get shuttled to that restaurant or that spot that everyone visits and posts on socials, but in traveling that way, you will miss all of the other amazing food, shops, and sights along the way that you might encounter by walking instead. Reading the docs is more like exploring the random nooks and crannies and finding experiences you weren't expecting and ultimately knowing more about the place you visited than if you had only visited the major tourist destinations.
As a senior-dev, I have generally a good idea of what to ask for because I have built many systems and learned many things along the way. A junior dev? They may not know what to ask for and therefore, may never discover those "detours" that would yield additional insights to tuck into the manifolds of their brains for future reference. For the junior dev, it's like the only trip they will experience is one where they just go to the well known tourist traps instead of exploring and discovering.
It's possible those arguments are correct. I wouldn't give up Google and SO, but I suspect I was learning faster when my first stop was K&R or a man page. There's a lot of benefit in building your own library of knowledge instead of cribbing from someone else's.
Of course no-one's stopping a junior from doing it the old way, but no-one's teaching them they can, either.
The naturally curious will remain naturally curious and be rewarded for it, everyone else will always take the shortest path offered to complete the task.
> The naturally curious will remain naturally curious and be rewarded for it
Maybe. The naturally curious will also typically be slower to arrive at a solution due to their curiosity and interest in making certain they have all the facts.
If everyone else is racing ahead, will the slowpokes be rewarded for their comprehension or punished for their poor metrics?
The thing is you need both. You need to have periods where you are reading through the docs learning random things and just expanding your knowledge, but the time to do that is not when you are trying to work out how to get a string into the right byte format and saved in the database as a blob (or whatever it is). Documentation has always has lots of different uses and the one that gets you answers to direct questions has improved a bit but its not really reliable yet so you are still going to have to check it.
No, trying stuff out is the valuable process. How I search for information changed (dramatically) in the last 20 years I've been programming. My intuition about how programs work is still relevant - you'll still see graybeards saying "there's a paper from 70s talking about that" for every "new" fad in programming, and they are usually right.
So if AI gets you iterating faster and testing your assumptions/hypothesis I would say that's a net win. If you're just begging it to solve the problem for you with different wording - then yeah you are reducing yourself to a shitty LLM proxy.
Yes, it is. And yes, it absolutely is harmful.
I think if this were true, then individualized mastery learning wouldn't prove to be so effective
https://en.wikipedia.org/wiki/Mastery_learning
Except none of us have a master teaching and verifying our knowledge on how to use a library. And AI doesn’t do that either.
No :)
Any task has “core difficulty” and “incidental difficulty”. Struggling with docs is incidental difficulty, it’s a tax on energy and focus.
Your argument is an argument against the use of Google or StackOverflow.
Not really. There’s a pattern to reading docs, just like there’s a pattern to reading code. Once you grasped it, your speed increase a lot. The slowness that junior has is a lack of understanding.
Complaining about docs is like complaining about why research article is not written like elementary school textbooks.
Disagree. While documentation is often out of date, the threshold for maintaining it properly has been lowered, so your team should be doing everything it can to surface effective docs to devs and AIs looking for them. This, in turn, also lowers the barrier to writing good docs since your team's exposure to good docs increases.
If you read great books all the time, you will find yourself more skilled at identifying good versus bad writing.
It really depends on what's being learned. For example, take writing scripts based on the AWS SDK. The APIs documentation is gigantic (and poorly designed, as it takes ages to load the documentation of each entry), and one uses only a tiny fraction of the APIs. I don't find "learning to find the right APIs" a valuable knowledge; rather, I find "learning to design a (small) program/script starting from a basic example" valuable, since I waste less time in menial tasks (ie. textual search).
> It really depends on what's being learned.
Also the difference between using it to find information versus delegating executive-function.
I'm afraid there will be a portion of workers who crutch heavily on "Now what do I do next, Robot Soulmate?"
Why?
If you can just get to the answer immediately, what’s the value of the struggle?
Research isn’t time coding. So it’s not making the developer less familiar with the code base she’s responsible for. Which is the usual worry with AI.
Struggling with poorly organized docs seems entirely like incidental complexity to me. Good learning resources can be both faster and better pedagogically. (How good today's LLM-based chat tools are is a totally separate question.)
Nobody said anything about poorly organized docs. Reading well structured and organized complex material is immensely difficult. Anyone who’s read Hegel can attest to that.
And yet I wouldn’t trust a single word coming out of the mouth of someone who couldn’t understand Hegel so they read an AI summary instead.
There is value in struggling through difficult things.
The best part is when the AI just makes up the docs
Feel free to waste your time sifting through a dozen wrong answers. Meanwhile the rest of us can get the answers, absorb the right information quickly then move on to solving more problems.
And you will have learned nothing in the process. Congratulations, you are now behind your peer who "wasted his time" but actually knows stuff which he can lean on in the future.
1965: learning how to punch your own punch cards is part of the learning process
1995: struggling with docs and learning how and where to find the answers part of the learning process
2005: struggling with stackoverflow and learning how to find answers to questions that others have asked before quickly is part of the learning process
2015: using search to find answers is part of the learning process
2025: using AI to get answers is part of the learning process
...
Has the quality of software been improving all this time?
This is both anachronistic and wrong.
To the extent that learning to punch your own punch cards was useful, it was because you needed to understand the kinds of failures that would occur if the punch cards weren't punched properly. However, this was never really a big part of programming, and often it was off-loaded to people other than the programmers.
In 1995, most of the struggling with the docs was because the docs were of poor quality. Some people did publish decent documentation, either in books or digitally. The Microsoft KB articles were helpfully available on CD-ROM, for those without an internet connection, and were quite easy to reference.
Stack Overflow did not exist in 2005, and it was very much born from an environment in which search engines were in use. You could swap your 2005 and 2015 entries, and it would be more accurate.
No comment on your 2025 entry.
> To the extent that learning to punch your own punch cards was useful, it was because you needed to understand the kinds of failures that would occur if the punch cards weren't punched properly. However, this was never really a big part of programming, and often it was off-loaded to people other than the programmers.
I thought all computer scientists heard about Dijkstra making this claim at one time in their careers. I guess I was wrong? Here is the context:
> A famous computer scientist, Edsger Dijkstra, did complain about interactive terminals, essentially favoring the disciplined approach required by punch cards and batch processing.
> While many programmers embraced the interactivity and immediate feedback of terminals, Dijkstra argued that the "trial and error" approach fostered by interactive systems led to sloppy thinking and poor program design. He believed that the batch processing environment, which necessitated careful, error-free coding before submission, instilled the discipline necessary for writing robust, well-thought-out code.
> "On the Cruelty of Really Teaching Computing Science" (EWD 1036) (1988 lecture/essay)
Seriously, the laments I hear now have been the same in my entire career as a computer scientist. Let's just look toward to 2035 where someone on HN will complain some old way of doing things is better than the new way because its harder and wearing hair shirts is good for building character.
Unironically, yes.
Now get back to work.
If the docs are poorly written then your not learning anything except how to control frustration
For an experienced engineer, working out the syntax, APIs, type issues, understanding errors, etc is the easy part of the job. Larger picture issues are the real task.
But for many Jr engineers it’s the hard part. They are not (yet) expected to be responsible for the larger issues.
what is a larger issue? lacking domain knowledge? or lacking deeper understanding of years of shit in the codebase that seniors may have better understanding? where I work, there is no issue that it "too large" for a junior to take on, it is the only way that "junior" becomes "non-junior" - by doing, not by delegating to so-called seniors (I am one of them)
"Larger issue" is overall technical direction and architecture, making decisions that don't paint you into a corner, establishing maintainability as a practice, designing work around an organization's structure and habit and so on.
But these are the things people learn through experience and exposure, and I still think AI can help by at least condensing the numerous books out there around technology leadership into some useful summaries.
You can't give a junior tasks that require experience and nuance that have been acquired over years of development. If you babysit them, then perhaps but then what is the point? By it's nature "nuance" is something hard to describe concretely but as someone who has mentored a fair few juniors most of them don't have it. AI generally doesn't have it either. Juniors need tasks at the boundary of their capability, but not far beyond to be able to progress. Simply allowing them to make a mess of a difficult project is not a good way to get there.
There is such a thing as software engineering skill and it is not domain knowledge, nor knowledge of a specific codebase. It is good taste, an abstract ability to create/identify good solutions to a difficult problem.
Unnecessary complexity, completely arbitrary one off designs, over emphasis on one part of the behavior while ignoring others. Using design patterns where they shouldn't be used, code once and forget operations exist, using languages and framework that are familiar but unfit for that purpose. The list goes on and I see it happen all the time, AI only makes it worse because it tend to verify all of these with "You're absolutely correct!".
Good luck maintaining that.
And this is always my question: "... because the genie, used well, accelerates learning." Does it though?
How are we defining "learning" here? The example I like to use is that a student who "learns" what a square root is, can calculate the square root of a number on a simple 4 function calculator (x, ÷, +, -) if iteratively. Whereas the student who "learns" that the √ key gives them the square root, is "stuck" when presented with a 4 function calculator. So did they 'learn' faster when the "genie" surfaced a key that gave them the answer? Or did they just become more dependent on the "genie" to do the work required of them?
You still need to be curious. I learn a ton by asking questions of the LLMs when I see new things. “Explain this to me - I get X but why did you do Y?”
It’s diamond age and a half - you just need to continue to be curious and perhaps slow your shipping speed sometimes to make sure you budget time for learning as well.
I think that's the "used well" in "because the genie, used well, accelerates learning".
We had 3 interns this past summer - with AI I would say they were VERY capable of generating results quickly. Some of the code and assumptions were not great, but it did help us push out some releases quickly to alleviate customer issues. So there is a tradeoff with juniors. May help quickly get features out, may also need some refactoring later.
Interesting how similar this is to the tradeoff of using AI coding agents
I think the big win with AI is being able to work around jargon. Don't know what that word means ask AI. what the history on it no problem. don't understand a concepts explain this at a high school reading level.
*Some juniors have gotten better.
I hate to be so negative, but one of the biggest problems junior engineers face is that they don't know how to make sense of or prioritize the gluttony of new-to-them information to make decisions. It's not helpful to have an AI reduce the search space because they still can't narrow down the last step effectively (or possibly independently).
There are junior engineers who seem to inherently have this skill. They might still be poor in finding all necessary information, but when they do, they can make the final, critical decision. Now, with AI, they've largely eliminated the search problem so they can focus more on the decision making.
The problem is it's extremely hard to identify who is what type. It's also something that senior level devs have generally figured out.
The cynic in me sees it as using juniors as a vehicle for driving up AI metrics. The seniors will be less critical reviewing AI output with a human shield/messenger.
Search is easily the best feature of AI/LLMs.
I kind of agree here. The mental model that works for me is "search results passed through a rock tumbler". Search results without attribution and mixed-and-matched across reputable and non-reputable sources, with a bias toward whatever source type is more common.
Which was trained on a pre-AI internet. What's going to happen in coming years when new tech comes out but perhaps isn't documented the same way anymore? It's not an unsolvable problem, but we could see unintended consequences like, say where you must pay the AI provider to ingest your data. Similar to buying poll space or AdSense or whatever they call it for search engines
If you release a new piece of technology from 2025 onwards and don't invest a decent amount of effort into producing LLM-friendly documentation (with good examples) that a user can slurp into their coding agent you're doing your new technology a disservice.
That's arguably all it ever was. Generating content using AI is just finding a point in latent space.
>but because the genie, used well, accelerates learning.
This is "the kids will use the AI to learn and understand" level of cope
no, the kids will copy and paste the solution then go back to their preferred dopamine dispenser
I've learned a lot of shit while getting AI to give me the answers, because I wanted to understand why it did what it did. It saves me a lot of time trying to fix things that would have never worked, so I can just spend time analyzing success.
There might be value in learning from failure, but my guess is that there's more value in learning from success, and if the LLM doesn't need me to succeed my time is better spent pushing into territory where it fails so I can add real value.
>I've learned a lot of shit while getting AI to give me the answers
I would argue you're learning less than you might believe. Similarly to how people don't learn math by watching others solve problems, you're not going to learn to become a better engineer/problem solver by reading the output of ChatGPT.
Do you honestly think that’s how people learn?
This is an example of a book on Common Lisp
https://gigamonkeys.com/book/practical-a-simple-database
What you usually do is follow the book instructions and get some result, then go to do some exploration on your own. There’s no walk in the dark trying to figure your own path.
Once you learn what works, and what does not, then you’ll have a solid foundation to tackle more complex subject. That’s the benefit of having a good book and/or a good teacher to guide you to the path of mastering. Using a slot machine is more tortuous than that.
I don't find it to be more torturous than that. In fact, if I were to go back and learn lisp again, I think I'd be a lot more motivated seeing how to build something interesting out of the gate rather than the toy programs I learned in my racket course.
Also, for a lot of things, that is how people learn because there aren't good textbooks available.
Define interesting.
I was helping a few people on getting started with an Android Development bootcamp and just being able to run the default example and get their bearing around the IDE was interesting to them. And I remember when I was first learning python. Just doing basic variable declaration and arithmetic was interesting. Same with learning C and being able to write tic-tac-toe.
I think a lot of harm is being done by making beginner have expectations that would befit people that have years of experience. Like you can learn docker in 2 months to someone that doesn't even know Linux exists or have never encountered the word POSIX.
Please do read the following article: https://www.norvig.com/21-days.html
Right, and they won't get hired beyond their internship.
Don’t confuse this with this persons ability to hide their instincts. He is redefining “senior” roles as junior, but words are meaningless in a world of numbers. The $$$ translation is that something that was worth $2 should now be worth $1.
Because that makes the most business sense.
I disagree. In my experience AI does most of the work and the juniors already poor skills atrophy. Then a senior engineer has to review AI slop and tell the junior to roll the AI dice again.
first response from me "let me mention how the real business world actually works" .. let's add a more nuanced slice to that however
Since desktop computers became popular, there have been thousands of small to mid-size companies that could benefit from software systems.. A thousand thousand "consultants" marched off to their nearest accountant, retailer, small manufacturer or attorney office, to show off the new desktop software and claim ability to make new, custom solutions.
We know now, this did not work out for a lot of small to mid-size business and/or consultants. Few could build a custom database application that is "good enough" .. not for lack of trying.. but pace of platforms, competitive features, stupid attention getting features.. all of that, outpaced small consultants .. the result is giant consolidation of basic Office software, not thousands of small systems custom built for small companies.
What now, in 2025? "junior" devs do what? design and build? no. Cookie-cutter procedures at AWS lock-in services far, far outpace small and interesting designs of software.. Automation of AWS actions is going to be very much in demand.. is that a "junior dev" ? or what?
This is a niche insight and not claiming to be the whole story.. but.. ps- insert your own story with "phones" instead of desktop software for another angle
One thing I'd point out is that there are only so many ways to write a document or build a spreadsheet. There are a ton of business processes that are custom enough to that org that they have to decide to go custom, change their process, or deal with the inefficiency of not having a technical solution that accomplishes the goal easily.
Lotus Notes is an example of that custom software niche that took off and spawned a successful consulting ecosystem around it too.
> Lotus Notes is an example
TIL Notes is still a thing. I had thought it was dead and gone some time ago.
I'm a little confused by this analysis. Are you saying that all enterprise software has been replaced with MS word and AWS?
certainly no -- not "all software" of anything. Where is the word "enterprise" in the post you have replied to ? "enterprise" means the very largest companies and institutions..
I did not write "all software" or "enterprise software" but you are surprised I said that... hmmm
I'm not swayed by appeals to authority, but this is a supremely bad take.
"AI" tools are most useful in the hands of experienced developers, not juniors. It's seniors who have the knowledge and capability to review the generated output, and decide whether the code is bullshit and will only cause more issues when it's merged, or whether it's usable if they tweak and adapt it in certain ways.
A junior developer has no such skills. Their only approach will be to run the code, test whether it fulfills the requirements, and, if they're thorough, try to understand and test it to the best of their abilities. Chances are that because they're pressured to deliver as quickly as possible to impress their colleagues and managers, they'll just accept whatever working solution the tool produces the first time.
This makes "AI" in the hands of junior developers extremely dangerous. Companies who allow this type of development will quickly grind to a halt under the weight of technical debt, and a minefield of bugs they won't know how to maneuver around.
The sad reality is that with "AI" there is no pathway for junior developers to become senior. Most people will gravitate towards using these tools as a crutch for quickly generating software, and not as a learning tool to improve their own skills. This should concern everyone vested in the future of this industry.
The amount of copium in the replies to this is just amazing. It’s amazing.
How would a person who describes himself as a "full time content producer" know what is actually going on in the industry?
https://substack.com/@kentbeck
What software projects is he actively working on?
The dude literally invented Extreme Programming and was the first signer of the Agile Manifesto. He's forgotten more about software development than most people on this site ever knew.
Someone's accomplishments don't make them incapable of having bad opinions and being wrong. Cults of personality are harmful to progress. Opinions should hold the same weight and be held to the same scrutiny regardless of who voiced them.
That wasn't the question being asked. The question being asked was literally "what are this guy's accomplishments," and Kent Beck is a tech industry OG with a laundry list of them.
Of course he can be wrong; he's human. That wasn't my point.
https://en.wikipedia.org/wiki/Kent_Beck
To be fair, even if I appreciate Beck, some people do get too famous and start to inhabit a space that is far removed from the average company. Many of these guys tend to give out advice that is applicable to a handful of top earning companies but not the rest.
Doesn't this back up the point? From his wiki it seems like he is mostly famous as a programming influencer, not as a programmer.
So? His bio is literally one latest fad after the other. Now he joins the "AI" fad, what a surprise.
That is a very cynical take which completely ignores his contributions through the decades.
In many cases he helped build the bandwagons you're implying he simply jumped onto.
> In many cases he helped build the bandwagons you're implying he simply jumped onto.
The fact that I cannot tell if you mean this satirically or not (though I want to believe you do!) is alarming to me.
Interesting take... I'm seeing a pattern... People think AI can do it all... BUT I see juniors often are the ones who actually understand AI tools better than seniors... That's what AWS CEO points out... He said juniors are usually the most experienced with AI tools, so cutting them makes no sense... He also mentioned they are usually the least expensive, so there's little cost saving... AND he warned that without a talent pipeline you break the future of your org... As someone who mentors juniors, I've seen them use AI to accelerate their learning... They ask the right questions, iterate quickly, and share what they find with the rest of us... Seniors rely on old workflows and sometimes struggle to adopt new tools... HOWEVER the AI isn't writing your culture or understanding your product context... You still need people who grow into that... So I'm not worried about AI replacing juniors... I'm more worried about companies killing their own future talent pipeline... Let the genies help, but don't throw away your apprentices.
ON TOP OF IT ALL, juniors are the ones who bring novel tools to the desk MOST times...i.e. I had no clue the Google IDE gave you free unlimited credits despite the terrible UI...but a young engineer told me about it!!
I've seen seniors and juniors bring novel tools in. Seniors do it less often perhaps - but only because we have seen this before under a different name and realize it isn't novel. (sometimes the time has finally come, sometimes it fail again for the same reason it failed last time)
I've seen seniors bring up novel ideas to juniors—well, novel to the juniors anyway.
Just an example. I've been in so many code bases over the years… I had a newer engineer come aboard who, when he saw some code I recently wrote with labels (!) he kind of blanched. He thought "goto == BAD". (We're talking "C" here.)
But this was code that dealt with Apple's CoreFoundation. More or less every call to CF can fail (which means returning NULL in the CF API). And (relevant) passing NULL to a CF call, like when trying to append a CF object to a CF array, was a hard crash. CF does no param checking. (Why, that would slow it down—you, dear reader, are to do all the sanity checking.)
So you might have code similar to:
CFDictionary dict = NULL;
dict = CFCreateDictionary();
if (!dict)
You would likely continue to create arrays, etc—insert them into your dictionary, maybe return the dictionary at the end. And again, you checked for NULL at every call to CF, goto bail if needed.Down past 'bail' you could CFRelease() all the non-null instances that you do not return. This was how we collected our own garbage. :-)
In any event, goto labels made the code cleaner: your NULL-checking if-statements did not have to nest crazy deep.
The new engineer admitted surprise that there might be a place for labels. (Or, you know, CF could have been more NULL-tolerant and simply exited gracefully.)
I'm just shocked people aren't clueing into the fact that tech companies are trying to build developer dependence on these things to secure a "rent" revenue stream. But hey, what do I know. It's just cloud hyper scaling all over again. Don't buy and drive your own hardware. Rent ours! Look, we built the metering and everything!
you will be able to run code assistants on your own machine soon, just like how you can run intense graphical simulations (videogames?) thanks to GPUs
idk about that with the rising cost of hardware. But I guess if your IT dept is doing the purchasing thats not really your problem.
I'm clued in to that but at this point who cares. All the models are fungible for the coding assistance use case.
I'd hope people are. It's painfully obvious this entire AI push is rent-seeking half hidden by big tech money. At some point the free money is going to go away, but the rent for every service will remain.
"BUT I see juniors often are the ones who actually understand AI tools better than seniors"
Sorry, what does that mean exactly ? Are you claiming that a junior dev knows how to ask the right prompts better than a Senior dev ?
Their implication is that junior devs have more likely built up their workflow around the use of AI tooling, likely because if they're younger they'll have had more plasticity in their process to adapt AI tooling
Overall I don't quite agree. Personally this applies to me, I've been using vim for the last decade so any AI tooling that wants me to run some electron app is a non starter. But many of my senior peers coming from VS Code have no such barriers
All the major players offer a CLI, for what it’s worth.
You won’t need Vim except to review changes and tweak some things if you feel like it.
Speaking of vim - adding and configuring copilot plugin for vim is easy (it runs a nodejs app in the background but if you have spare 500 Mb RAM it's invisible).
Copilot in vim is not the same as cursor. e.g. There is no multiline tab autocomplete.
Speaking as a senior dev, anecdotally juniors may indeed understand AI tools better, because they spend more hours a day coding and working with the tools, and they need the tools to understand the codebase or to be productive. Seniors have more hours stuck in meetings, developing specs/tickets for the juniors, code reviewing, etc. Seniors are likely to not bother with a prompt for simple changes in codebases they already understand.
Some old dogs resist learning new tricks.
If AI is just prompts to you, you fall into the "don't know how to use it" group.
Amazon has an internal platform for building software. The workflows are documented and have checks and balances. So the CEO wants to have more junior developer that are more proficient with AI, and have (in ratio) less senior developers. Also, product context comes from (product) managers, UX designers.
For medium or small companies, these guardrails or documentation can be missing. In that case you need experienced people to help out.
> I'm seeing a pattern...
Me too. Fire your senior devs. (Ha ha, not ha ha.)
No no, fire them.
Cannot wait for the 'Oh dear god everything is on fire, where is the senior dev?' return pay packages.
Sorry but what the heck is up with all the ellipses in this comment?
They have an emacs package that triples their . automatically!
It's a sort of stream of consciousness. That style of writing goes in and out of style from time to time but some people use it consistently.
They're trying really hard to make sure you know they didn't write their post with an LLM? /s
honestly i think that'll be a thing in the future
"bespoke, hand generated content straight to your best readers"
Nah, models can be fine tuned and trained on anything. Common consumer products like ChatGPT and Gemini have particular styles, very polite and helpful, but there are models trained to be combatative, models trained to write in the style of shakespeare, all sorts of things. Someone could train a model to reply to posts in the style of HN comments and you’d probably never know.
7/10 senior devs (usually fellas 50+) will get mad at you for trying to use Claude Code. Me: “dude it writes better code than crap you write in your mush middle-age brain.” Also me: “I also have mush brain.”
I think LLM is a reflection of human intelligence. If we humans become dumber as a result of LLM, LLMs will also become dumber. I’d like to think in some dystopian world, LLM’s trained from pre 2023 data will be sought after.
is this just a janky summary cause you added zero new viewpoints
you're right but my opinion about this has changed
I would have agreed with you 100% one year ago. Basically senior engineers are too complacent to look at AI tools as well as ego driven about it, all while corporate policy disincentivizes them from using anything at all, with maybe a forced Co-Pilot subscription. While junior engineers will take a risk that the corporate monitoring of cloud AI tools isn't that robust.
But now, although many of those organizations are still the same - with more contrived Co-Pilot subscriptions - I think senior engineers are skirting corporate policy too and become more familiar with tools.
I'm also currently in an organization that is a total free for all with as many AI coding and usage tools as necessary to deliver faster. So I could be out of touch already.
Perhaps more complacent firms are the same as they were a year ago.
Maybe, but you make it sound like juniors are more worthy to companies than seniors. Then fire most/all seniors and good luck with resulting situation.
Coding in any sufficiently large organization is never the main part of senior's time spend, unless its some code sweatshop. Juniors can do little to no of all that remaining glue that makes projects go from a quick brainstorming meeting to live well functioning and supported product.
So as for worth - companies can, in non-idedal fashion obviously, work without juniors. I can't imagine them working without seniors, unless its some sort of quick churn of CRUDs or eshops from some templates.
Also there is this little topic that resonates recently across various research - knowledge gained fast via llms is a shallow one, doesn't last that long and doesn't go deeper. One example out of many - any time I had to do some more sophisticated regex-based processing I dived deep into specs, implementation etc. and few times pushed it to the limits (or realized task is beyond what regex can do), instead of just given the result, copypasted it and moved along since some basic test succeeded. Spread this approach across many other complex topics. That's also a view on long term future of companies.
I get what you say and I agree partially but its a double edged sword.
Btw, it's funny that certain kind of people tend to believe in common sense only when it comes from the mouth of the AWS CEO or a similar persona.
I - senior - can patch an application in an unknown language and framework with the AI. I know enough to tell it to stop the wildly stupid ideas.
But I don't learn. That's not what I'm trying to do- I'm trying to fix the bug. Hmm.
I'm pretty sure AI is going to lead us to a deskilling crash.
Food for thought.
I think the temptation to use AI is so strong that it will be those who will keep learning who will be valuable in future. Maybe by asking AI to explain/teach instead of asking for solution direclty. Or not using AI at all.
On the other hand, if it's a one-off, you'll have forgotten what you learned by the time you'd need to use that skill again.
> I'm pretty sure AI is going to lead us to a deskilling crash.
That's my thought too. It's going to be a triple whammy
1. Most developers (Junior and Senior) will be drawn in by the temptation of "let the AI do the work", leading to less experience in the workforce in the long term.
2. Students will be tempted to use AI to do their homework, resulting in new grads who don't know anything. I have observed this happen first hand.
3. AI-generated (slop) code will start to pollute Github and other sources used for future LLM training, resulting in a quality collapse.
I'm hoping that we can avoid the collapse somehow, but I don't see a way to stop it.
I think seniors know enough to tell whether they need to learn or not. At least that's what I tell myself!
The thing with juniors is: those who are interested in how stuff works now have tools to help them learn in ways we never did.
And then it's the same as before: some hires will care and improve, others won't. I'm sure that many juniors will be happy to just churn out slop, but the stars will be motivated on their own to build deeper understanding.
On the contrary, being able to access (largely/verifiably) correct solutions to tangible & relevant problems is an extremely great way to learn by example.
It should probably be supplemented with some good old RTFM, but it does get us somewhat beyond the "blind leading the blind" StackOverflow paradigm of most software engineering.
I can't help but feel this is backpedaling after the AI hype led to people entering university avoiding computer science or those already in changing their major. Ultimately we might end up with a shortage of developers again, which would be amusing.
I went to university 2005-2008 and I was advised by many people at the time to not go into computer science. The reasoning was that outsourced software developers in low-cost regions like India and SEA would destroy salaries, and software developers should not expect to make more than $50k/year due to the competition.
Even more recently we had this with radiologists, a profession that was supposed to be crushed by deep learning and neural networks. A quick Google search says an average radiologist in the US currently makes between $340,000 to $500,000 per year.
This might be the professional/career version of "buy when there's blood in the streets."
You nailed it on the head, down to the exact examples.
I was still in high school in 2010, and was told the same thing about outsourcing to India/SEA/etc. making a CS degree/career (in the US) a terrible choice. It wasn't just random people saying this either, I was reading about it in the news, online, had some family acquaintances with alleged former software dev career, etc. I didn't listen, and I am glad I didn't.
As I was graduating from college, and deep learning was becoming a new hot thing, I heard the same thing about radiologists, and how they are all getting automated away in the next 5 years. I had no plans to go to med school, and I didn't know anyone at the time who went through it, so I didn't know much about the topic. On the surface, it seemed like a legitimate take, and I just stored it in my head as "sounds about right."
Cue to now, I know more than a few people who went through med school, and am in general more attuned to the market. Turns out, all of that was just another genpop hype, those news articles about "omg radiologists are all getting replaced by computers" stopped from showing up on any of my news feeds, and not a single radiology-specialized med school graduate I know had any issues with getting a job (that paid significantly more than an entry level position at a FAANG).
I have zero idea what point I was trying to make with this comment, but your examples mirror my personal experience with the topic really well.
I went for CS in my late 20s, always tinkered with computers but didn't get into programming earlier. College advisor told me the same thing, and that he went for CS and it was worthless. This was 2012.
I had a job lined up before graduating. Now make high salary for the area, work remotely 98% of the time and have flexible schedule. I'm so glad I didn't listen to that guy.
The one thing I learned in college is that the advisors are worthless. There's how many students? And you are supposed to expect they know the best thing for you? My advisor told me that all incoming freshmen must take a specific math class, a pre-calculus course, totally ignoring all of my AP exams that showed I was well beyond that. Wasted my time and money.
My take is that these are not binary issues. With outsourcing, it is true that you can hire someone cheaper in Asian countries but it cannot kill all jobs locally. So what happens is that the absolute average/mediocre get replaced by outsourcing and now with AI while the top talent can still command a good salary because they are worth it.
So I think that a lot of juniors WILL get replaced by AI not because they are junior necessarily but because a lot of them won't be able to add great value compared to a default AI and companies care about getting the best value from their workers. A junior who understands this and does more than the bare minimum will stand out while the rest will get replaced.
> So I think that a lot of juniors WILL get replaced by AI not because they are junior necessarily but because a lot of them won't be able to add great value compared to a default AI and companies care about getting the best value from their workers. A junior who understands this and does more than the bare minimum will stand out while the rest will get replaced.
Again this is what people said about outsourced developers. 2008 logic was, why would anyone hire a junior for $50k/year when you could hire a senior with 20 years experience for $10k/year from India?
Reality: for 5+ years you could change careers by taking a 3-6 month JavaScript bootcamp and coming out the other end with a $150k job lined up. That's just how in demand software development was.
If you have to be “top talent” to survive it’s not a good field to get into anymore.
To survive against Outsourcing/cheaper labor and AI, I would agree.
> Even more recently we had this with radiologists, a profession that was supposed to be crushed by deep learning and neural networks. A quick Google search says an average radiologist in the US currently makes between $340,000 to $500,000 per year.
At the end of the day, radiologists are still doctors.
Yep, the only reason their pay is high is artificial barriers to entry.
Yup hearing big talk about competition and doom is a strong signal that there is plenty of demand.
You can either bet on the new unproven thing claiming to change things overnight, or just do the existing thing that's working right now. Even if the new thing succeeds, an overnight success is even more unrealistic. The insight you gain in the meantime is valuable for you to take advantage of what that change brings. You win either way.
When there is no competition that is a sign there is no demand.
There can sometimes be too much competition, but often there is only the illusion of too much if you don't look at quality. You can find a lot of cheap engineers in India, but if you want a good quality product you will have to pay a lot more.
Can anyone really blame the students? If I were in their shoes, I probably wouldn't bother studying CS right now. From their perspective, it doesn't really matter whether AI is bullshit in any capacity; it matters whether businesses who are buying the AI hype are going to hire you or not.
Hell, I should probably be studying how to be a carpenter given the level at which companies are pushing vibe coding on their engineers.
"after the AI hype led to people entering university avoiding computer science or those already in changing their major"
That's such a terrible trend.
Reminds me of my peers back in ~2001 who opted not to take a computer science degree even though they loved programming because they thought all the software engineering jobs would be outsourced to countries like India and there wouldn't be any career opportunities for them. A very expensive mistake!
Certainly, I even know of experienced devs switching out of tech entirely. I think the next couple of decades are going to be very good for software engineers. There will be an explosion of demand yet a contraction in supply. We are in 2010 again.
There will be programmers of the old ways, but AI is basically code 2.0, there are now a lot of things that are AI specific that those with traditional software development skills can’t do.
Like what exactly?
It's backpedaling but I don't think it's planning ahead to prevent a developer shortage - rather it's pandering to the market's increasing skepticism around AI and that ultimately the promised moonshot of AI obsoleting all knowledge work didn't actually arrive (at least not in the near future).
It's similar to all those people who were hyping up blockchain/crypto/NFTs/web3 as the future, and now that it all came to pass they adapted to the next grift (currently it's AI). He is now toning down his messaging in preparation of a cooldown of the AI hype to appear rational and relevant to whatever comes next.
"We were against this all along"
The party line will be: “we always advised using it if it as long as it helps productivity.”
Pointing out that it wasn’t always that will make you seem “negative.”
You are right, perfect amount of false humility and balance. The wage suppression is an accidental biproduct and not the intent. Collateral damage if you will.
Or maybe they realize the AI needs humans in the loop for the foreseeable future for enterprise use cases and juniors (and people from LCL areas) are cheaper and make the economics make some sort of sense.
Agreed.
Considering the talk around junior devs lately on HN, there's way too many of them, it would indeed be amusing.
> changing their major
To what?
So he's saying we should be replacing the seniors with fresh grads who are good at using AI tools? Not a surprising take, given Amazon's turnover rate.
My experience is that juniors have an easier time to ramp up, but never get better at proper engineering (analysis) and development processes (debug). They also struggle to read and review code.
I fear that unless you heavily invest in them and follow them, they might be condemned to have decades of junior experience.
> but never get better at proper engineering (analysis) and development processes (debug). They also struggle to read and review code.
You can describe pre-ai developers and like that too. It's probably my biggest complaint about some of my Co workers
I have the same experience.
In my view there's two parts to learning, creation and taste, and both need to be balanced to make progress. Creation is, in essence, the process of forming pathways that enable you to do things, developing taste is the process of pruning and refining pathways to doing things better.
You can't become a chef without cooking, and you can't become a great one without cultivating a taste (pun intended) for what works and what it means for something to be good.
From interactions with our interns and new-grads, they lack the taste, and rely too much on the AI for generation. The consequence is that when you have conversations with them, they straggle to understand the concepts and tools they are using because they lack the familiarity that comes with creation, and they lack the skills to refine the produced code into something good.
A Junior Dev is not just for Christmas, it’s for life.
But more seriously are there CEOs out there who think they can replace the people starting off in their industry with AI? Who do the think will be the senior devs in 5-10yrs?
Meanwhile:
"Amazon announces $35 billion investment in India by 2030 to advance AI innovation, create jobs" https://www.aboutamazon.com/news/company-news/amazon-35-bill... (Dec 9 2025)
oh so CEO is probably referring to using AI to replace foreign junior devs is a bad idea. Domestic junior dev replacement is ok though
This is for data locality.
Juniors are also more likely to be the MOST proficient/comfortable with AI tooling.
Pair them with a senior so they can learn engineering best practices:
And now you've also just given your senior engineers some extra experience/insights into how to more effectively leverage AI.
It accelerates the org to have juniors (really: a good mix of all experience levels)
> In fact, 30% of companies that laid off workers expecting savings ended up increasing expenses, and many had to rehire later.
Such as (cough...) Amazon?
He said the same thing four months ago: https://news.ycombinator.com/item?id=44972151
We frequently get juniors or interns who are perfectly capable of pumping out many LoC with the use of AI in various forms - the issue is that they _don't_ actually ever learn how to think for themselves, and can't fix problems when something goes wrong or the LLM paints itself into a corner. I have found myself doing a lot more shepherding and pairing with juniors when they can't figure something out recently, because they just have not had the space to build their own skills.
1. replacing junior engineers, with AI ofcourse breaks the talent pipeline. Seniors will retire one day, who is going to replace them? Are we taking the bet, that we wont need any engineer at that time? sounds dangerous.
2. Junior engineer's heavy reliance on AI tools is a problem in itself. AI tools learn from existing code that is written by senior engineers. Too much use of AI by junior engineers will result in deterioration of engineering skills. It will eventually result in AI learning from AI generated code. This is true for most other content as well, as more and more content on internet is AI generated.
I believe the idea is to not stop hiring juniors. Instead it's to replace anybody that commands a high salary with a team of cheaper juniors armed with LLM's. The idea is more about dragging down average pay than never hiring anybody. At least for now.
And then all those unemployed seniors with extensive domain knowledge use AI to speedrun the creation of competition and you need to spend $$$$ to buy them out and shut them down. Solid idea.
Junior vs senior is the wrong framing. It's "can use LLMs effectively" vs "can't use LLMs effectively".
It's like expecting someone to know how to use source control (which at some point wasn't table stakes like it is today).
I have heard this thing quite a few times over last few months each time is Amazon or AWS CEOs. May be this time he want to replace senior engineers. That would be more useful for them as each passing year they more and more of them and in times like these they are not looking to go leave Amazon on their own.
Well, yeah. Then who will become the senior engineers in 10-15 years?
You think the people deciding whether to hire more juniors are planning more than one or two quarters ahead? 10-15 years is someone else's problem.
Yes, unfortunately 10-15 years is 5 sequential someones else's problems.
These people are working on destroying the planet to make more money, they absolutely do not care. Our society isn't set up to punish them, but encourage such behavior to even more extremes (see datacenter build outs causing water shortages, electricity hikes, and cancer in poor communities; nearly every politician capitulating on such actions because they don't know better).
I wish people would get off the "AI is the worst thing for the environment" bandwagon. AI and data centers as a whole aren't even in the top 100 emitters of pollution and never will be.
If you want to complain about tech companies ruining the environment, look towards policies that force people to come into the office. Pointless commutes are far, far worse for the environment than all data centers combined.
Complaining about the environmental impact of AI is like plastic manufacturers putting recycling labels on plastic that is inherently not recycleable and making it seem like plastic pollution is every day people's fault for not recycling enough.
AI's impact on the environment is so tiny it's comparable to a rounding error when held up against the output of say, global shipping or air travel.
Why don't people get this upset at airport expansions? They're vastly worse.
Of course they aren't polluters as in generating some kind of smoke themselves. But they do consume megawatts upon megawatts of power that has to be generated somewhere. Not often you have the luxury of building near nuclear power plant. And in the end you're still releasing those megawatts as heat into the atmosphere.
> Why don't people get this upset at airport expansions?
We do too, don't worry.
Pretty sure Anthropic is hoping the answer is Claude.
Pretty sure Antropic knows their hopes won't come true. They just won't tell you that.
Obviously after 10-15 years of experience working as a developer AI will be a senior dev. Probably will get promoted to management with all that experience.
Promoting your best engineers to management sometimes gets you a great manager, but often gets you a mediocre or just-about-competent manager at the cost of a great engineer.
I'm a big fan of the "staff engineer" track as a way to avoid this problem. Your 10-15 year engineers who don't vibe with management should be able to continue earning managerial salaries and having the biggest impact possible.
https://staffeng.com/about/
I'm also a fan of leadership without management. Those experienced engineers should absolutely be taking on leadership responsibilities - helping guide the organization, helping coach others, helping build better processes. But they shouldn't be stuck in management tasks like running 1-1s and looking after direct reports and spending a month every year on the annual review process.
This is a general problem that corporations have trouble with with: The struggle to separate leadership and people management. Why does the person who tells you what to do also need to be the same person who does your annual review, who also has to be the same person who leads the technical design of the project, approves your vacation, assists with your career development, and gives feedback or disciplinary correction when you mess up? Why do we always seem to bundle all these distinct roles together under "Manager"?
This is exactly where I find myself. I've been asked several times to take on management, but I have no interest in it. I got to be a principal after 18 years of experience by being good at engineering, not management. Like you said, I can and do help with leadership through mentorship, offering guidance and advice, giving presentations on technical topics, and leading technical projects.
Absolutely agree. Regardless, my org keeps trying to get me to take a management role after 15 years dev experience. I love my job and don't like managing people. You couldn't pay me enough to become a manager.
I still spend a week on annual reviews but you make great points all around.
10-15 years? Sadly, most people are declared senior after like 2 years of work.
Anything besides next quarter does not exist.
You can’t pretend you know where technology will take us.
AI will be the senior engineer
Stolen from your competitors, obviously.
I gave opus an "incorrect" research task (using this slash command[1]) in my REST server to research to use SQLite + Litestream VFS can be used to create read-replicas for REST service itself. This is obviously a dangerous use of VFS[2] and a system like sqlite in general(stale reads and isolation wise speaking). Ofc it happily went ahead and used Django's DB router feature to implement `allow_relation` to return true if `obj._state.db` was a `replica` or `default` master db.
Now claude had access to this[2] link and it got the daya in the research prompt using web-searcher. But that's not the point. Any Junior worth their salt — distributed systems 101 — would know _what_ was obvious, failure to pay attention to the _right_ thing. While there are ideas on prompt optimization out there [3][4], the issue is how many tokens can it burn to think about these things and come up with optimal prompt and corrections to it is a very hard problem to solve.
[1] https://github.com/humanlayer/humanlayer/blob/main/.claude/c... [2] https://litestream.io/guides/vfs/#when-to-use-the-vfs [3] https://docs.boundaryml.com/guide/baml-advanced/prompt-optim... [4]https://github.com/gepa-ai/gepa
I'm not sure a junior would immediately understand the risks of what you described. Even if they did well in dist sys 101 last year.
Same story from 4 months ago: https://news.ycombinator.com/item?id=44972151
Ive been managing and supporting teams for a long time and i'm sorry, but junior and mid-level devs do the majority of the heavy lifting when it comes to work output in big corps. I don't think AI will replace them. I don't think all these IC5 and IC6 engineers are going to be putting up 400-500 diffs a year anytime soon.
I recently pair-worked with two junior developers (on their first job, but still with like 2+ years with the company) in order to transfer the know-how of something.
I realized that they are shockingly bad at most basic things. Still their PR:s look really good (on the surface). I assume they use AI to write most of the code.
What they do excel in is a) cultural fit for the company and b) providing long-term context to the AIs for what needs to be done. They are essentially human filters between product/customers and the AI. They QA the AI models' output (to some extent).
> A company that relies solely on AI to handle tasks without training new talent could find itself short of people.
I kind of agree with this point from the perspective of civilisation.
Which is a less dumb idea: replacing new grad junior devs with AI or H1Bs?
Better idea, replace everyone with AI, and when that doesn't work, replace the AI with H1Bs.
What about new grad H1Bs?
Hire new grads on OPT and transition to H-1B.
But I think the actual reason was not addressed. The work of junior devs is exactly what can be replaced by AI, instead of the more complex abilities senior development possess.
so what is the take away message? fire only the senior devs cause they cost too much and can't use AI?
old zuckerburg said 80% of junior developers would be cut in 2026, I say 80% of CEO's who replace their software engineers will be cut in 2026
The third point is applicable to general demographic survival as well. Countries with most child birth rates, would ultimately win.
The level of cynicism here is astronomical. After discovering the strategy of "fire juniors and let a few seniors manage autonomous agents" was an abject failure, now the line is "actually juniors are great because we've brainwashed them into thinking AI is cool and we don't have to pay them so much". Which makes me want to vomit.
The only relevant point here is keeping a talent pipeline going, because well duh. That it even needs to be said like it's some sort of clever revelation is just another indication of the level of stupid our industry is grappling with.
The. Bubble. Cannot. Burst. Soon. Enough!!
> AWS CEO says using AI to replace junior staff is 'Dumbest thing I've ever heard' (theregister.com) 1697 points by JustExAWS 3 months ago
https://news.ycombinator.com/item?id=44972151
Does this story add anything new?
This CEO is a sales guy. He's gonna repeat same thing every few months claiming "Brand new and improved AI Prediction"
Somebody gets it.
some companies doing the opposite, firing senior devs and hiring junior with AI experience
Am I going crazy or he already said that several months ago?
Now with AI, I expect junior developers to learn much quicker and progress to senior very quickly. I'd now rather hire at least 1 of each to begin with, both "junior" and a "senior" developer and then additionally hire more juniors to quickly turn them into a "senior".
We do not need to hire anymore outside senior developers who need to be trained on the codebase with AI, given that the junior developers catch up so quickly they already replaced the need to hire a senior developer.
Therefore replacing them with AI agents was quite premature if not completely silly. In fact it makes more sense to hire far less senior developers and to instead turn juniors directly into senior developers to save lots of money and time to onboard.
Problem solved.
If only everyone thought this way.
on a separate note- If AI eats the SaaS, what will happen to AWS?
Umm.. more business. Because AI runs in cloud and consume even more resources.
but which business?
4) Junior devs have an incomparably superior context window.
5) less attitude
To me the more insidious problem is that we have juniors now that aren’t learning much because they lean on AI for everything. They are behind the curve.
…proceeds to replace junior devs with AI
They know AI is inefficient and mostly just a glorified "template filler" at this point...
This is performative bullshit pandering to the increased skepticism around AI. He wouldn't be saying that if AI investment was still in full swing.
I do agree with him about AI being a boon to juniors and pragmatic usage of AI is an improvement in productivity, but that's not news, it's been obvious since the very beginnings of LLMs.
So it's performative when the head of AWS says it and not news. But it's not performative when you say it and people should have listened to you in the comments?
It's performative when you talk whatever the market wants to hear rather than sticking to an opinion (no matter how flawed it is). This behavior reminds me of the cryptobros that were hailing NFTs/web3 as the next best thing since sliced bread, and when that didn't came to pass quietly moved onto the next grift (AI) with the same playbook.
(also I’m just talking out of my ass on a tech forum under a pseudonym instead of going to well-publicized interviews)
FINALLY.
shareholder wouldn't like this
They must’ve forgot who created the first tech hype bubbles in the first place bc I’m about to replace some of these companies if I don’t get hired soon
Most of the apps that I use regularly fail at least once a day nowadays. I think this is a direct cause of putting AI code in production without reviewing/QA.
While I have no particular love for AI generated code, I think this has nothing to do with AI. Software has been unreliable for over a decade. Companies have been rushing out half baked products and performing continual patches for many years. And it's our fault because we have just come to accept it.
> Software has been unreliable for over a decade
The "over" deserves a lot of emphasis. To this day, I save my code at least once per line that I type because of the daily (sometimes hourly) full machine crashes I experienced in the 80s and 90s.
Same, I think I should just turn on autosave at this point to save my fingers
My prediction is that this will actually get better, because the cost to find and fix with AI is so much lower in time investment.
The problem is human, not technical. Companies and managers need to start caring about the details instead of crossing items off a list. Until we see that culture shift in the industry, which might never happen, AI isn't going to help—if anything, it'll make the problem worse as devs rush to deliver on arbitrary deadlines.
Well the reason I think it might be different is that I am noticing a material change in my behavior.
I have always cared a lot about quality and craftsmanship. Now when I am working and notice something wrong, I just fix it. I can code it entirely with AI in the time it would've take me to put it on an eternal backlog somewhere.
Plus, if you are skipping tests or telling yourself, you wrote them when they don’t actually verify anything in the first place, then buying into a hype cycle of “the AI writes perfect code“ is unlikely to break the pattern
AI doesn't ever fix anything, it just breaks stuff and adds technical debt.
This sounds like a comment from someone who doesn't have visibility into how good the models are getting and how close they are to fully autonomous, production-grade software development.
This is an easy theory to prove; if AI was anywhere close to a senior engineer, we'd see the costs of software development drop by a corresponding amount or quality would be going up. Not to mention delivery would become faster. With LLMs being accessible to the general public I'd also expect to see this in the open-source world.
I see none of that happening - software quality is actually in freefall (but AI is not to blame here, this began even before the LLM era), delivery doesn't seem to be any faster (not a surprise - writing code has basically never been the bottleneck and the push to shove AI everywhere probably slows down delivery across the board) nor cheaper (all the money spent on misguided AI initiatives actually costs more).
It is a super easy bet to take with money - software development is still a big industry and if you legitimately believe AI will do 90% of a senior engineer you can start a consultancy, undercut everyone else and pocket the difference. I haven’t heard of any long-term success stories with this approach so far.
This sounds like a comment from someone who has tested it in a limited capacity such as small blog sites or side projects that did not need to be maintained
Yes, you are right
I'm yet to see that production-grade code written by these production-grade models;