Touching on the atrophy point, I actually wrote a few thoughts about this yesterday: https://www.neilwithdata.com/outsourced-thinking

I actually disagree with Andrej here re: "Generation (writing code) and discrimination (reading code) are different capabilities in the brain." and I would argue that the only reason he can read code fluently, find issues, etc. is because he has spent year in a non-AI assisted world writing code. As time goes on, he will become substantially worse.

This also bodes incredibly poorly for the next generation, who will mostly in their formative years now avoid writing code and thus fail to even develop a idea of what good code is, how it works/why it works, why you make certain decisions, and not others, etc. and ultimately you will see them become utterly dependent on AI, unable to make progress without it.

IMO outsourcing thinking is going to have incredibly negative consequences for the world at large.

Thanks, this rings true to me. The struggle is an investment, and it pays off in good judgement and taste. The same goes for individual codebases too. When I see some weird bug and can immediately guess what’s going wrong and why, that’s my time spent in that codebase paying off. I guess LLM-ing a feature is the inverse, incurring some kind of cognitive debt.

Read your blog post and agree with some of it. Largely I agree with the premise that the 2nd and 3rd order effects of this technology will be more impactful than the 1st order “I was able to code this app I wouldn’t have otherwise even attempted to”. But they are so hard to predict!

Is coding like piloting, where pilots need a certain number of hours of "flight time" to gain skills, and then a certain number of additional hours each year to maintain their skills? Do developers need to schedule in a certain number of "manually written lines of code" every year?