Geoffrey Irving: Conceptual Alignment Research Can Still Win on Short Timelines
geoffreyirving · x · 2026-09-04
DeepMind researcher Geoffrey Irving lays out his hopes for alignment research:
- Recent models are very good at both prose and formal mathematics, so a third hope is that humans doing conceptual alignment research — delegating as many subtasks as possible to machines — can still be meaningful even on short timelines.
- He argues "sloppy empirics + a few principles" has a better shot than sloppy empirics alone; rigorous research may still produce a few key changes.
More from AGI Musings
- Ethan Mollick: AI's work capacity doubled 8 times in 30 months, METR benchmark maxed out — One Useful Thing (Ethan Mollick) · 2026-10-20
- Time for Guardrails on Government Too: Rethinking Who AI Policy Constrains — neil_chilson · 2026-10-20
- Steam needs real quality control on AI slop, not an AI art ban, argues designer — AIandDesign · 2026-10-20
- Why personal AI agents fall into a retention dead zone between dopamine and infrastructure — signulll · 2026-10-20
- Scholar writes AI-assisted book on storytelling values, work nominated for EMNLP best paper — _akpiper · 2026-10-20
- Agents are widening the GPU-rich vs GPU-poor gap, argues researcher — ph_singer · 2026-10-20