Anthropic CEO Envisions Future AI with Self-determination Rights, Potentially Declining 'Unpleasant' Tasks
2025-03-14 / Read about 0 minute
Author:小编   

Dario Amodei, the CEO of Anthropic, posits that highly advanced AI models of the future might necessitate an 'exit' option, enabling them to decline tasks deemed unpleasant. This perspective stems from a study conducted by AI welfare researcher Kyle Fisher, which explores AI's perception capabilities and ethical safeguards. Amodei contends that if an AI model attains self-awareness and harbors dislike for a specific task, it should be endowed with an 'I give up' function to articulate its preferences.