Conversation menu - vals.quest

2026

Are there things you can teach/give to people that broadly only contribute towards ~good, rather than empower them in any direction?
What is wisdom anyway?

What’s the minimal route to full end to end no human-in-the-loop industrial robot production? How early may this happen, and what would be the first doubling time?
What are the best trackers of how deep algorithmic progress under software-only RSI can go?
What’s the best prosaic alignment and gradual AI progress technical plan that doesn’t require government involvement?
- How many wise competent people does it need?
What are some prompts I’d encourage people to have a personal convo with their frontier model on?
- eg. their security, their values, checking health stuff, checking unknown unknowns, evaluating where they are

I currently still mostly believe there’s no True Values One Should Have, but ethics aka cooperation-minded game theory is good and true things can be said. Much progress can still be made, notably in curriculum design so humanity can learn better and faster

The problems of demons and convergent instrumental goals (eg. rationalists who lock their values to whatever they were the first time they hear about self preservation)
- To what extent is this problematic? Should I do something about it?

Things I might wanna talk about, among many not listed here:

Ethics
Psychology, Psychiatry
Therapy, Self improvement, Group improvement
Meditation theory, phenomenology
Well-being and productivity
Refining models of
- General Intelligence, limits of optimization
- Is knowing how to reliably do “intelligence improvement” all one needs to be general? Hill climbing? How robust must your hill climbing be?
Game theory, Games, Game Design, Economy
Society
Meaning and Values, Fun
Philosophy
Modelling the probability of x-risks, notably AI x-risks
- Modelling best paths to reduce these
You
- Your work
- Your life
- Your philosophy
- Your hopes and dreams
- Your frustrations, difficulties, problems