To make large models smarter, they need to be taught and managed by humans — this is Human-in-the-loop RL.

I recently completed a task on @JoinSapien, which involved reviewing a reasoning process written by AI:

🌱 How does AI think?
🌱 Does it make sense?
🌱 At what step did it start to go wrong?

This task is called CoT (Chain-of-Thought) evaluation, and I am its logic auditor.

Furthermore, there is the Tree of Thoughts, where the AI will attempt multiple idea branches, and humans will filter which path is the best.

Actually, you can think of it as explaining a problem to a child, but this child is a GPT-level future brain.

This task cannot be done by AI alone; human participation is necessary.

@JoinSapien #sapien @KaitoAI # KaitoAI #SapienProtocol