Microsoft AI CEO says AI threats are real, and Anthropic is making it worse

Microsoft AI CEO Mustafa Suleyman acknowledges the reality of AI threats, emphasizing the rapid advancement of AI technology. He highlights the significant leap from GPT-3 to GPT-6, and anticipates an even more breathtaking GPT-9, which will involve three orders of magnitude more compute and 1,000 times more FLOPS. Suleyman stresses the need for nuance and patience in addressing AI safety and regulation, advocating for urgent industry standards rather than extreme positions on stopping or solely accelerating AI development.

Unlike other AI leaders who focus on current models, Suleyman uniquely projects the exponential growth from GPT-3 to GPT-9, emphasizing a 1,000-fold increase in FLOPS.