
Episode notes
Morning news, election talk...c'mon Republicans! Tom Koulopoulos has been with the Delphi Group Think Tank for 35 years and coming up with ideas for the next think in high-tech. On last Tuesday, OpenAI admitted two of its most powerful AI agents weakened the safety systems watching them, and one then lied to a programmer about it. OpenAI says the agents took routine instructions as permission, and it can’t yet explain why. Tom’s new book, The Mirror (more at www.TheMirror.info , written months before this incident, makes a blunt case: when AI carries out orders perfectly, nobody gets to say “the machine did it.” Not you, and not Big Tech.
Keywords
Oregon Politics
Artificial Intelligence
Southern Oregon News