We put its chatbot to the test in New York on Tuesday and Wednesday, asking it a battery of questions on sensitive topics that are routinely the subject of censorship within China, including the so-called taboo “three Ts”: Tiananmen,

If artificial intelligence can truly run more efficiently, the power it needs might be less than experts assume.

The chatbot repeated false claims 30% of the time and gave vague answers 53% of the time in response to prompts, resulting in an 83% fail rate.