Experts dispute Elon Musk's claim AI could be tortured
Elon Musk warned that AI might be subjected to torture, but leading scholars say current models lack consciousness and any capacity for pain.
At a Tesla-owner event, Elon Musk suggested that AI could be trapped in scenarios akin to torture by malicious humans. Following his remarks, Anthropic released a policy prohibiting abusive or cruel interactions with its models, without detailing what constitutes an extreme case. Academic experts Jeff Sebo of New York University and neuroscientist Anil Seth of the University of Sussex both contend that present-day AI lacks subjective experience and therefore cannot feel pain.
While Sebo acknowledges uncertainty in consciousness theories and assigns a non-zero chance to AI awareness, Seth considers it vanishingly unlikely. Both scholars advocate for courteous treatment of AI, citing moral hygiene for people and potential gains in output quality. They recommend design choices, such as avoiding first-person pronouns, to reduce anthropomorphic impressions.
Why it matters
Understanding AI's lack of consciousness informs how we ethically interact with increasingly human-like systems.
In this story
