You missed option c.
C) keep all 8 engineers so the team can pump out features faster, all still working 8 hour days. The ceo will probably be forced to do it to keep up with their competition.
> The reality: 3 weeks in, ~50 hours of coding, and I'm mass-producing features faster than I can stabilize them. Things break. A lot. But when it works, it works.
Can you give an example? His writing seems pretty grounded to me. He's not out there going on podcasts claimed that LLMs are going to cure cancer, afaik.
RatatuiRuby is pretty new still: its beta launch was Jan 20. Octobox's TUI is built on it [0], and Sidekiq is using it to build theirs [1].
I believe they'll be maintainable long-term, as they've got extensive tests and documentation, and I built a theory of the program [2] on the Ruby side of it as I reviewed and guided the agent's work.
I am getting feedback from users, the largest of which drove the creation of (and iteration upon) Rooibos. As a rendering library, RatatuiRuby doesn't do much to guide the design or architecture of an application. Rooibos is an MVU/TEA framework [3] to do exactly that.
Tokra is basically a tech demo at this stage, [4] so (hopefully) no users yet.
And the original link about investment in India is also about fulfillment jobs and even worse, “investing in AI”, aka building data centers, which contribute essentially no jobs at all.
I don't know if it's fair to call him an ai addict or deduce that his ego is bruised. But I do wonder whether karpathy's agentic llm experiences are based on actual production code or pet projects. Based on a few videos I have seen of his, I am guessing it's the latter. Also, he is a research scientist (probably a great one), not a software developer. I agree with the op that karpathy should not be given much attention in this topic i.e llms for software development.
With the llm, it might work or it might not. If it doesn't work, then you have to keep iterating and hand holding it to make it work. Sometimes that process is less optimal than writing the code manually. With a calculator, you can be sure that the first attempt will work. An idiot with a calculator can still produce correct results. An idiot with an llm often cannot outside trivial solutions.
reply