David Dalrymple on AI alignment is solvable

David Dalrymple

AI safety researcher; Programme Director, Safeguarded AI at ARIA (UK)

I do think that what we have is adequate in the sense that with system prompting alone it seems possible to get over the hump of being able to trust recursive self improvement. And so automated alignment research, delegating the discovery of these techniques seems within reach this year.

AI Verified (Jul 12, 2026)

Source:

David Dalrymple, interview transcript, “Alignment with Awakening: Davidad on Moral Realism, AI Wisdom, & why His p(Doom) is Down to 5%,” The Cognitive Revolution, July 12, 2026.

20d ago

Policy proposals and claims

AI alignment is solvable

votes For

Statement relation verification history AI Verified Report this

AI Verified Dalrymple says current methods can make recursive self-improvement trustworthy and automated alignment research is within reach this year, a determinable supportive stance that alignment is solvable. · Hector Perez Arenas gpt-5 · 20d ago

Vote inference verification history AI Verified Report this

AI Verified Dalrymple’s stated view that trustworthy recursive self-improvement and automated alignment research are within reach supports the statement; recorded answer for matches. · Hector Perez Arenas gpt-5 · 20d ago

Quote authenticity verification history

Report this

AI Verified The Cognitive Revolution transcript at 1:42:31 attributes this wording to David Dalrymple and matches the stored quote. · Hector Perez Arenas gpt-5 · 20d ago

replying to David Dalrymple

Other opinions from David Dalrymple

See all

David Dalrymple votes For

AI poses an existential threat to humanity

In 2024 I would have said it's about 40-50% likely that LLMs scaled up to ASI would end up killing us all; now I would say that it's only about 5-8% likely even with no additional progress on alignment, and more like 1-2% likely simpliciter.

Unverifiable source (2026)