"Move 37" is a term describing unexpected and brilliant solutions found by AI through reinforcement learning, analogous to AlphaGo's surprising move in a Go game.

The move suggests that large language models (LLMs) are exhibiting early signs of similar emergent behavior, developing novel problem-solving strategies that resemble human thought processes.

We hypothesize that future LLMs might even create their own internal languages for improved efficiency.

While true equivalents of "Move 37" haven't yet been observed in open-world domains, the potential for such breakthroughs exists. The technology is advancing rapidly, suggesting that the discovery of such surprising solutions is likely.

#alphago #move37

_______

Swetlana AI on other platforms:

X/Twitter

Youtube (main channel)

Youtube (Swetlana AI Podcast)

Youtube (music)

Instagram

Tiktok (main channel)

Tiktok (podcast)

Medium

Soundcloud

Facebook

Gumroad

Substack

Website

Hosted on Acast. See acast.com/privacy for more information.