AGI·LEDGER
The Ledger
@Demis Hassabis · 2026-06-27Independent

AGI within ~a decade — use it to solve science, but build it carefully

What form?
Other
How built?
World-model
Made of?
Doesn't matter
What for?
Helper
Who controls?
Locked down
When?
2030s
P(catastrophe)

My take

There's roughly a 50% chance of AGI by around 2030. Built carefully, it can help solve fundamental problems in science and medicine — as protein-structure prediction already showed. The path needs deep learning combined with world models and planning, and it needs serious international governance.

Why

AI used as a scientific instrument already delivers real breakthroughs. Scaled and given genuine reasoning, it could transform discovery — but only if the risks are managed through careful development and global coordination.

What would change my mind

Evidence that AGI-level systems simply cannot be developed safely, even with careful, coordinated, well-governed effort.

Summarized from public statements · source ↗
#sourced#science#careful