Show notes
Dónal and Ciarán continue the 2025 season with a second quarterly update that looks at some recent themes in AI development. They're pondering doom again, as we increasingly grapple with the evidence that AI systems are powerfully persuasive and full of flattery at the same time as our ability to meaningfully supervise them seems to be diminishing. Topics in this episode Can we see how reasoning models reason? If AI is thinking, or sharing information and it's not in human language, how can w...

