Think you can spot a chatbot in a conversation? Think again. A recent study reveals that OpenAI’s GPT-4.5, one of the leading AI models, is so sophisticated that it fools people into thinking it’s human more often than actual humans do in a modern take on the classic Turing test.
The Turing test, a long-standing benchmark for machine intelligence and human perception, was recently revisited by researchers at UC San Diego’s Language and Cognition Lab. In this contemporary version, nearly 300 participants put their skills to the test.
Participants engaged in simultaneous conversations with both a human and an AI. Their task was simple, yet challenging: determine which conversationalist was the human and which was the AI. The results proved quite revealing about how far AI has come.
