The Age of Agents: Can multiple LLM agents improve outputs?

This was a quick experiment on creating multiple LLM (GPT) agents with their own objective to see if outputs can be improved.

As in, a consensus network? IMO, yes. Is that scary? Yes.