Episode Details

AI safety pioneer Roman Yampolskiy argues that frontier AI could become impossible to switch off—and Brian Keating presses him on what that means for the future. This condensed version turns the full Into The Impossible conversation into a fast listen, distilling the key ideas in minutes. You’ll hear why Yampolskiy believes superintelligent systems could outpace human oversight, why he calls many current guardrails “safety theater,” and why he thinks pausing frontier model development may be the only rational path if control can’t be proven. He also discusses AI alignment, machine learning, job automation, regulation, productivity tools, and the limits of human decision-making when systems move far faster than we can verify. If you want the core argument on AI ethics, control, and the coming labor shake-up, this summary gives you the essentials fast. Listen now to get the key ideas in minutes.