AIPhys.org Technologyabout 2 hours ago
AIs Fail to Detect Other AIs in Tests
5 min read
Seven large language models failed to reliably distinguish humans from other AIs in a new study. Researchers from Paris Cité University conducted 238 tests involving 1,714 questions. AI evaluators identified AI participants as human 88% of the time, compared to 78% for actual humans. The study concludes current AI screening is insufficient for ensuring authenticity in dialogue.
Level
Hype check
Tap to vote and see what everyone thinks.
