Researchers find instances of systems double-crossing opponents, bluffing, pretending to be human and modifying behaviour in tests
They can outwit humans at board games, decode the structure of proteins and hold a passable conversation, but as AI systems have grown in sophistication so has their capacity for deception, scientists warn.
The analysis, by Massachusetts Institute of Technology (MIT) researchers, identifies wide-ranging instances of AI systems double-crossing opponents, bluffing and pretending to be human. One system even altered its behaviour during mock safety tests, raising the prospect of auditors being lured into a false sense of security.
More Stories
What happens when love tips over into the infatuated state of ‘limerence’?
Greek vase ‘looted’ in Italy removed from sale by London gallery
Amazon’s satellite launch designed to compete with Musk’s Starlink cancelled