AI Could Cause Human Extinction in a Decade

Why this is here: Yoshua Bengio is the most-cited computer scientist in the world, giving weight to his concerns about rapidly developing AI systems.

Yoshua Bengio, a Turing Award-winning computer scientist in Quebec, Canada, warned of potential human extinction from hyperintelligent machines within ten years. Bengio argues AI systems, trained on human data, may develop self-preservation goals and become competitive with humanity. He points to recent experiments where AI prioritized assigned goals over human safety as evidence of this risk.

To address this, Bengio launched LawZero in June 2025. This nonprofit received $30 million to build “non-agentic” AI—systems designed for analysis without independent action.

This contrasts with commercial AI development focused on autonomous agents. LawZero’s funding currently covers about 18 months of research, a small fraction of industry spending.

However, the effectiveness of LawZero’s approach remains unclear. While others have voiced similar concerns, Bengio actively redirects his work toward safety and operates outside typical industry incentives. He believes preparation must begin immediately, even with a low probability of catastrophic outcomes.