Week 5 · The character of what we build

What should an AI do when values collide?

AI dispositions can influence advice, institutions and public reasoning. Examine honesty, deference, care and independence through concrete dilemmas, then choose a small project you can finish.

Current character training may not transfer to more capable systems. Evaluations can reward persuasive language rather than reliable conduct.

What result would persuade you that your preferred character trait creates a risk?

Write two pre-proposals with a question, four-week output, neglectedness case and personal fit.

Interactive tools load with JavaScript. Original sources retain their language and attribution.