All that money, we are not even sure we can build AGI. What is AGI. Clearly scaling LLMs won't cut it, but VCs keep funding people because they pretend they can build super intelligence. I don't see that happening in the next 5 years: https://medium.com/@fsndzomga/there-will-be-no-agi-d9be9af44...
What’s your evidence that scaling won’t improve AI?
Chess is a pretty good example. You could theoretically train an LLM on just chess games. The problem is there are more chess positions than atoms in the universe. So you can’t actually do it in practice. And chess is a much more constrained environment than life. At any chess position there are only ~35 moves on average. Life has tons of long-tail situations which have never been seen before.
And for chess we already have superhuman intelligence. It doesn’t require trillion-dollar training clusters, you can run a superhuman chess bot on your phone. So there are clear questions of optimality as well: VC money should be aware of the opportunity cost in investing money under “infinite scaling” assumptions.