U.S. Markets Plunge Amid Tariff Turmoil

A recent study reveals that OpenAI's GPT-4.5 model has surpassed the Turing Test, successfully convincing human evaluators that it is a person 73% of the time. This achievement signals a significant advancement in artificial intelligence, traditionally measured by the Turing Test—developed in 1950 by mathematician Alan Turing. According to Cameron Jones, a researcher at UC San Diego's Language and Cognition Lab and lead author of the study, the results indicate that the AI could effectively impersonate humans in brief interactions.
The research involved nearly 300 participants, with some acting as interrogators and others as either human or chatbot “witnesses.” GPT-4.5 performed better when prompted to adopt a persona, achieving the impressive 73% success rate, compared to only 36% when instructed to engage without a specific identity. Jones warns that such advancements could lead to job automation and increased risks in social engineering attacks.
Social media reactions reflect a mix of amusement and concern, with some users questioning the implications of machines that may be better at mimicking human behavior than humans themselves.