How Science Can Fix Dishonest AI - Yoshua Bengio with Nicholas Thompson

About this episode

Yoshua Bengio discusses the potential for a new AI model that prioritizes accuracy and honesty over pleasing users.

In the years following the launch of ChatGPT, as concerns spread over the social and political impacts of LLMs, one person’s warnings seemed particularly dire: Yoshua Bengio’s, a scientist and one of the “godfathers” of AI. The potential negative impacts of his life’s work weighed so heavily on Bengio that he signed his name to an open letter advocating for a pause in AI research. (The pause didn’t happen.) But recently, Bengio has found renewed optimism as he pursues a project dubbed “Scientist AI.” The pitch: What if AI didn’t care about pleasing us, and instead, like a scientist, prioritized accuracy, honesty, and probable outcomes? In a conversation with Nicholas Thompson, CEO of The Atlantic, Bengio outlines why he thinks this approach will produce better outcomes, the challenges to implementing a model that polices other (often better-funded) models, and why the age of AI– so far marked by an international arms race– will need greater international cooperation. (00:00) Introduction (05:00) Can we understand what's happening inside neural network vectors and attention systems? (07:00) How ChatGPT changed Bengio’s risk assessment (09:37) The case for optimism (at least when…

People in this episode

Host: Nicholas Thompson

Guest: Yoshua Bengio

Topics covered

• AI ethics
• neural networks
• scientific approach to AI
• international cooperation
• alignment problem

Keywords

• AI
• Scientist AI
• ChatGPT
• neural networks
• alignment problem
• international cooperation
• accuracy
• honesty

Mentioned in this episode

Organizations: The Atlantic

Products: ChatGPT

More episodes of The Most Interesting Thing in AI

• How AI Is Changing Code - Paul Ford with Nicholas Thompson · June 10, 2026 · 49 min
• The AI Jobs Disruption - Erik Brynjolfsson with Nicholas Thompson · May 27, 2026 · 1h 2m
• The Limits of Predictive AI - Carissa Véliz with Nicholas Thompson · May 20, 2026 · 48 min
• AI Utopia or Catastrophe? Nick Bostrom with Nicholas Thompson · May 13, 2026 · 48 min
• The Case for Open-Source AI - with Nicholas Thompson and Raffi Krikorian · May 6, 2026 · 54 min
• Sam Altman on Where AI Models Go Next, with Nicholas Thompson · April 29, 2026 · 53 min