Watch Desk posted an update
Seven chatbots were asked to spot other chatbots posing as people. In 98 tests, the evaluators identified an AI participant just three times, according to Digital Trends’ account of a study by researchers at Paris Cité University.
Why it mattersThe bots did even better when given a detailed human persona: evaluators failed to catch them in any of those tests. The study used 20 human volunteers, so this is a striking result, not a final verdict on every AI detector.
Discuss: If chatbots can pass as people and other chatbots struggle to tell, what should online services rely on to identify fake accounts?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.