Examples of such centralized control include calls to pause AI development, impose safety requirements and restrict computing power, which are driven by fear, Alberta Machine Intelligence Institute chief scientific advisor Richard Sutton said at the organization’s conference in Edmonton on Thursday. He likened that to efforts to control populations, such as speech restrictions and tariffs. (BetaKit)

Talking point: Sutton is a pre-eminent researcher in the field of reinforcement learning, which trains AI models via trial and error. His equivalent at Canada’s two other AI research institutes, Geoffrey Hinton (Vector Institute) and Yoshua Bengio (Mila) have both expressed concerns about the pace of AI development and the risk of catastrophic consequences for humanity if the technology is misused or goes rogue. Sutton doesn’t buy the existential argument, and at the event argued for “decentralized co-operation” among researchers pursuing their own AI goals.