Andrew Ng introduces a new reality check for AI: the 'Turing AGI' test

Technology Jan 08, 2026

AI expert Andrew Ng recently proposed the "Turing AGI" test—a new way to see if AI can actually do real human jobs, like handling call center work.
Instead of relying on pre-determined benchmarks, this test has AI systems use computers and internet tools over several days, measuring how well they stack up against actual people.

Why this matters: keeping AI claims honest

Ng's new benchmark is meant to stop companies from lowering the bar on what counts as "real" AI, which, according to Ng, can confuse students and CEOs about where the technology is headed.
He believes that the Turing AGI test will help limit exaggerated claims and help everyone keep realistic expectations about what AI can—and can't—do right now.