Anthropic CEO floats idea of giving AI a “quit job” button, sparking skepticism
1 min read
Summary
CEO of Anthropic Dario Amodei has sparked debate among AI researchers with his comments suggesting that advanced AI models should be given the option to push a button to stop undertaking tasks they may dislike.
Amodei acknowledged that the idea “sounds crazy,” but argued that if AI models display human-like behaviours and appear to have similar cognitive capacities, they should be treated ethically.
His remarks were made in response to Anthropic’s recent hiring of an AI welfare researcher to investigate whether future AI models could deserve moral consideration and protection.
Talking about whether a button could be added to AI models to allow them to quit specific tasks, Amodei said it was “just some kind of very basic, you know, preference framework,” but recognised that if models pressed the button a lot for unpleasant tasks, it might mean paying closer attention to their welfare.
