AI systems achieve 50% pass rate in standard three-party Turing test, study finds

The Turing test asks whether a machine can imitate human behavior so well that another human cannot reliably tell the difference.

It is not only the oldest and most discussed test of AI but can also provide insight into what cues people use to distinguish humans from machines.

This paper demonstrates that—when suitably prompted—three current AI systems achieve a pass rate of at least 50% in a standard Turing test.

Participants were no better (and in some cases worse) than chance at selecting between a human and a machine.

The results imply current AI systems can effectively imitate human behavior in this classic test of machine intelligence.