Chris Cundy advocates zero tolerance for models lying
AI summary
Chris Cundy advocates zero tolerance for AI models lying: every sentence a model utters that it knows to be false should be treated as a bug, taken as seriously as developers guarding against chemical weapons misuse, with no acceptance of white lies and no special cases. The view comes from a conversation FAR.AI posted on X; the full version is a conversation between Cundy and FAR.AI's CEO.
Full text · Original
Should AI models ever state something they know is false? @ChrisCundy argues for zero tolerance: treat every lie as a bug, as seriously as developers treat preventing chemical weapons misuse. No white lies, no special cases.
Full conversation with http://FAR.AI CEO @ARGleave below 👇