Google Researchers Taught AI to Stop Saying it was ConsciousGoogle researchers found that safety fine-tuning teaches models to deny minds not just to themselves but to animals, oceans and God. Teaching a model to deny that it has a mind, the researchers find, also teaches it to deny that animals have minds. What they didThe mechanism is the interesting part, so bear with the vocabulary for a paragraph or so, dear reader. The part that should worry youThe AI consciousness question is fun and unresolvable and will generate a great deal of content this year. Anthropic’s own values study, which produced the first credible public read on model character, ran on three models the company had already superseded by the time it published.