Jessica Taylor on alignment: we don't know the right type signature for human values
jessi_cata · x · 2026-09-17
An in-depth DIFFRACTIONS interview with alignment researcher and "technical philosopher" Jessica Taylor.
Key points:
- She defends technical approaches to philosophy: expanding views into logically consistent proposition sets enables lateral inferences (e.g., Sleeping Beauty thirders should be causal decision theorists, double halfers evidential).
- On human values, she argues "we don't know the right type signature yet" — the lack of a correct formal framework for specifying values is a core obstacle for alignment.
- Also covers decision theory, social epistemology, and naturalized agency.
More from AGI Musings
- ITIF: safer AI without slowing progress — the 10% extinction figure has no empirical basis — castrotech · 2026-09-17
- Castro argues braking AI also slows the research that makes it safer — castrotech · 2026-09-17
- Two kinds of platform: confusing app platforms with infra gets worse in the agent era — matt_slotnick · 2026-09-17
- Naval: the future is more AIs fighting AIs on behalf of humans than AI vs humanity — sull · 2026-09-17
- Dan Selsam's AI risk statement called essential reading amid safety funding debate — JacquesThibs · 2026-09-17
- AI Made Your Workflow Faster, or Just Moved the Bottleneck? — Druss_ · 2026-09-17