[talldayo]

I think the assumption is flawed because the value generated by AI is spurious. The demand is there, and the supply hasn't yet run out, but people aren't rushing to rework their lives around AI. We haven't seen serious advancements that AI can spearhead besides non-LLM applications like computer vision. And those aren't new, they're decades-old.

To bring another analogy into the equation, AI isn't crude oil. No country depends on AI to operate, for defensive purposes or as a critical decision-making heuristic - it's a speculative market. AI is another test for businesses a-la cryptocurrency, probing the few last firms that claim to "innovate" to see what they can put out. So far they've done excellent on the marketing side, but just about nothing in terms of practical or valuable problem solving. In 5 years I feel pretty confident that LLMs will be treated like a sideshow attraction and not a practical tool used by earnest humans.