I was approaching this from political engagement and conflict resolution given the problems representative democracy throws up with the Principle-agent problem. So I would suggest rather than having the AI vote -we can go into reasons of why this can be a problem- I’d pair it with a open source transparent dedicated Conflict resolution LLM to mediate the process and provide decision options. It isn’t there to decide things rather provide the options and guardrails to what is feasible. Something like Getting to Yes looking at positions and interests not the actors themselves helps separate partisan takes. Also the work on Sway and argument is a great addition to the mix and definitely worth discussing.
Anyway this is something of interest to me I’m more theory than how to build anything. I also have an interest in a journalism LLM with adversarial agent Bayesian logic. Are we able to take this to email?
I was approaching this from political engagement and conflict resolution given the problems representative democracy throws up with the Principle-agent problem. So I would suggest rather than having the AI vote -we can go into reasons of why this can be a problem- I’d pair it with a open source transparent dedicated Conflict resolution LLM to mediate the process and provide decision options. It isn’t there to decide things rather provide the options and guardrails to what is feasible. Something like Getting to Yes looking at positions and interests not the actors themselves helps separate partisan takes. Also the work on Sway and argument is a great addition to the mix and definitely worth discussing.
Anyway this is something of interest to me I’m more theory than how to build anything. I also have an interest in a journalism LLM with adversarial agent Bayesian logic. Are we able to take this to email?