> I fired up ChatGPT and asked somewhat fancifully, “Why should I learn to add when a calculator can do it for me?”
And yet the prompt itself contains "play the devil's advocate for me” and they are surprised when the LLM replied "maybe we don't need to learn"?
While I applaud the author's transparency in showing the full chat log, I can't help but wonder if the general public being surprised by LLM capability is from the lack of imagination itself.
Every time I prompt an LLM, I start off with an expectation of what the output should look like, what keywords are likely to appear in the proceeding response. In this case, both "no, you should still learn" and "yes, we should offload lower level work to machine like compilers" are well within the expected output. Paired with the priming of "devil's advocate", is it any surprising that the response chose the latter result?
Writing all this out, maybe it's why I am yet to be able to get on the RSI doom-hype cycle.
The usage of the ChatGPT transcript was intended more as a literary device to capture the reader’s attention; it merely introduces one of the themes of the article, ChatGPT’s answer itself is not the focus. But thank you for reading! I appreciate your feedback.
The advent of AI is starting to make us seriously wonder whether we should continue to cultivate our own minds or just let a computer do the heavy lifting. In some ways, this question is not new but the answer does look a little different. I humbly present The Case for Learning, my take on why it’s important and you should still bother doing it.
Early on in the article:
> I fired up ChatGPT and asked somewhat fancifully, “Why should I learn to add when a calculator can do it for me?”
And yet the prompt itself contains "play the devil's advocate for me” and they are surprised when the LLM replied "maybe we don't need to learn"?
While I applaud the author's transparency in showing the full chat log, I can't help but wonder if the general public being surprised by LLM capability is from the lack of imagination itself.
Every time I prompt an LLM, I start off with an expectation of what the output should look like, what keywords are likely to appear in the proceeding response. In this case, both "no, you should still learn" and "yes, we should offload lower level work to machine like compilers" are well within the expected output. Paired with the priming of "devil's advocate", is it any surprising that the response chose the latter result?
Writing all this out, maybe it's why I am yet to be able to get on the RSI doom-hype cycle.
The usage of the ChatGPT transcript was intended more as a literary device to capture the reader’s attention; it merely introduces one of the themes of the article, ChatGPT’s answer itself is not the focus. But thank you for reading! I appreciate your feedback.
The advent of AI is starting to make us seriously wonder whether we should continue to cultivate our own minds or just let a computer do the heavy lifting. In some ways, this question is not new but the answer does look a little different. I humbly present The Case for Learning, my take on why it’s important and you should still bother doing it.
[flagged]