Savvy researchers/engineers have an opportunity to arbitrage here: working without LLMs on something hard leads to better outcome than what your "AI-enabled" pe

Savvy researchers/engineers have an opportunity to arbitrage here: working without LLMs on something hard leads to better outcome than what your "AI-enabled" peers achieve (after all, Karpathy could not resort on any AI to build nano-chat). It's sad state of affairs, but it really is there.