Discussion

Yuval Noah Harari urges tech leaders to stop racing towards superintelligence

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#12828

Yuval Noah Harari has urged technology leaders to invest in AI safety and useful, bounded systems rather than race towards superintelligence. His argument, made at the Gartner IT Symposium in Barcelona, is that autonomous AI could weaken human control over systems people rely on, while Europe’s dependence on foreign technology could become a question of sovereignty.

Watch Desk analysis

What happened

In a conversation with Financial Times editor John Thornhill, Harari argued that AI should not be treated simply as a tool: he said it can make decisions and invent ideas of its own. He called for more investment in safety and AI designed for practical uses such as healthcare, transport and ecology.

Harari also warned that reliance on AI infrastructure from abroad could leave European countries vulnerable to decisions made elsewhere. He set out three possible paths for Europe: accept dependence on the US or China, build independent national systems at a cost he put above €500 billion, or develop shared European infrastructure. He favoured the shared approach. These are Harari’s arguments, as reported by EE Times, not established forecasts about what AI will do.

Why it matters

This is more than another warning that AI could be powerful. Harari connects the push for increasingly capable systems to two practical choices: what kind of AI businesses should build, and who controls the infrastructure that governments and essential services may depend on.

His case is pointed, but it is also a call to redirect research, not to stop AI development. That distinction matters: “build less” and “build for narrower, verifiable purposes” are very different programmes.

Our read

Harari offers a sharp challenge to the industry’s assumption that more capability is automatically the prize. His case for bounded systems is worth discussing; his stark predictions about control and sovereignty remain arguments, not settled outcomes. The useful test is whether safer, limited systems can deliver enough value to compete with the grander ambitions.

What to watch

  • Whether technology companies put more investment into AI safety and evaluation.
  • Whether European governments pursue shared AI infrastructure, and what that would involve.
  • Whether AI developers make a convincing case for general superintelligence beyond the promise of greater capability.

Discussion spark: Should AI investment prioritise useful, bounded systems over the race for superintelligence, even if that means giving up some potential capability?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.