New AI Paradigm: Communitized RL

It's not a good moat. If we train an LLM using the same company's employees, that makes more sense.

AGI is nowhere near, vertical agents popping up in cyber and biology also signals this.