Thanks so much for this article, I feel like I understand LLMs and their potential trajectory far better than I did before. The framings of “drunk expert” vs. “interns” was a little bit of a lightbulb moment for me.
As a non-technical person I’ve tried reading a lot of theoretical stuff about LLMs and AI both here and on LessWrong. I usually start reading then at some point (in the first third/half of the article) I hit a point where I genuinely don’t understand what’s going on, stop reading then go back and read something else usually in my global health comfort zone :D.
For whatever reason I managed to read this whole article while at least feeling like I understood every paragraph and concept. I think the bar for understanding here is lower than many AI explainer articles, which I greatly appreciated and I would encourage other non-AI-technical people to read this in full. I would consider myself likely in the bottom 20% of forum readers when it comes to AI understanding as well.
I’m sure its not easy to walk the line between adequately explaining these concepts while still enabling basic people like me to understand ;).
Thanks so much for this article, I feel like I understand LLMs and their potential trajectory far better than I did before. The framings of “drunk expert” vs. “interns” was a little bit of a lightbulb moment for me.
As a non-technical person I’ve tried reading a lot of theoretical stuff about LLMs and AI both here and on LessWrong. I usually start reading then at some point (in the first third/half of the article) I hit a point where I genuinely don’t understand what’s going on, stop reading then go back and read something else usually in my global health comfort zone :D.
For whatever reason I managed to read this whole article while at least feeling like I understood every paragraph and concept. I think the bar for understanding here is lower than many AI explainer articles, which I greatly appreciated and I would encourage other non-AI-technical people to read this in full. I would consider myself likely in the bottom 20% of forum readers when it comes to AI understanding as well.
I’m sure its not easy to walk the line between adequately explaining these concepts while still enabling basic people like me to understand ;).
So well done.