Contrary to the idea that safety compromises capability, Bengio believes 'Scientist AI' could be even more capable. This is because it's trained to explicitly reason about statements and produce structured, decomposable chains of reasoning, similar to mathematical proofs. This structured approach, he suggests, could offer an advantage over current 'chain-of-thought' methods that may produce plausible but unverified outputs.

Impact: High. This challenges a common assumption in AI development, proposing that safety and enhanced capability can be achieved simultaneously. It suggests that a more rigorous, truth-oriented AI architecture might unlock new levels of performance.

In the source video, this keypoint occurs from 00:55:00 to 00:56:38.

Sources in support: Rob Wiblin (Host)