Comment by overgard
6 hours ago
It's a really underrated problem. I don't think my actual cognitive skills have declined by using AI, but I do notice that my patience and attention span are a lot lower.
I'm learning a new code base for a new job right now, and I'm finding AI to be a really double edged sword for it. One one hand, it's extremely valuable for asking questions about the code base. On the other hand, if I'm not careful and I just let it apply the fix before I even investigate it, I'm really not learning the code base well at all. I find I need to actually write new code in a code base to exercise the necessary mental muscles to actually retain understanding.
Incidentally, I do find that this large new code base I'm learning also shows the limitations of AI. There's no way I can vibe features on this without understanding and not introduce a lot of issues. Even targeted bug fixes have a lot of unintended consequences the LLM doesn't see. This isn't a bad code base at all, but it's definitely at the size where even frontier models struggle. So to me that tells me that the argument that I should just use more AI to solve my AI issues and not bother to understand the code base isn't viable at the moment.
> I don't think my actual cognitive skills have declined by using AI
I'm not speaking about you but... I know most people would not have much awareness of their cognitive decline. I know this because that awareness gap is there with or without LLMs, across all age groups and cultures.
Especially since attention -- which the parent commenter says has been diminished by LLMs -- is a key part of cognition.
adding to the irony is the fact that the mechanism used by the NN architecture in LLMs is also called attention
For cognition, sure, but that's a fairly weak claim. A dog that chases its tail for 3 hours might be considered conscious, but maybe not highly intelligent.
The attention deficit part of ADHD hurts some people, but a high intelligence is able to make up for it in other ways. Attention span is a multiplier for intelligence. Someone with a lower IQ but a longer attention span is able to outperform a higher IQ but shorter attention span, traditionally.
What's required though, is the attention span and the memory to really dig deep into a problem, and then go for a run. If AI makes that easier, since it lets you skip the boring parts and get to the meat of the problem, then hey.
True, I guess I try to have some objective measures like my chess elo and maybe some canaries like what books I'm reading. But it would be really hard to tell.