Geoffrey Irving on AI alignment is solvable

Geoffrey Irving

AI safety researcher; former safety researcher at OpenAI and Google DeepMind, former chief scientist at the UK AI Security Institute, and cofounder and chief scientist of Resolution.

I think that could work. I don’t think we have a strong argument that the pragmatic mixture of approaches will get all the way there, but it just seems very dicey, and our understanding of the dynamics involved is very weak.

AI Verified (Aug 11, 2026)

Source:

Geoffrey Irving, interview with Tom Reed, “#251 – Geoffrey Irving on how to solve alignment before superintelligence arrives,” 80,000 Hours, August 11, 2026.

20d ago

Policy proposals and claims

AI alignment is solvable

abstains

Statement relation verification history AI Verified Report this

AI Verified Irving directly discusses whether alignment approaches can succeed; his qualified uncertainty supplies a determinate stance. · Hector Perez Arenas gpt-5.6 · 19d ago

Vote inference verification history AI Verified Report this

AI Verified Irving describes a possible solution but emphasizes weak evidence and uncertainty; recorded abstain matches. · Hector Perez Arenas gpt-5.6 · 19d ago

Quote authenticity verification history

Report this

AI Verified 80,000 Hours transcript contains the exact Geoffrey Irving wording. · Hector Perez Arenas gpt-5.6 · 19d ago

replying to Geoffrey Irving

Other opinions from Geoffrey Irving

See all

Geoffrey Irving votes Against

Pausing AI at human-genius level to solve alignment is safer than either racing to superintelligence or halting entirely

Now. My take is that if we were to carefully analyse this question of exactly when we should slow down, it would be a while ago in the past, because we’re just too close to this crazy future. Trying to be super precise about exactly when in the futur...

more AI Verified source (Aug 11, 2026)

added 26d ago

Geoffrey Irving votes For

AGI could quickly lead to superintelligence

Yeah. I think the main uncertainty here is: are the models going to be good not just at verifiable tasks with clean rewards, but also fuzzier things — intuition, fuzzy planning, this kind of thing? I think people are overweighting the probability tha...

more AI Verified source (Jun 29, 2026)

added 21d ago