AI tools like GPT-4 are so advanced that distinguishing between human and machine conversation partners is becoming increasingly difficult. At UC San Diego, a controlled Turing Test was conducted with ELIZA, GPT-3.5, and GPT-4. Participants conversed with either a human or AI for five minutes and had to guess if their partner was human. GPT-4 was identified as human 54% of the time, outperforming GPT-3.5 and ELIZA. Participants were no better than chance at identifying GPT-4 as AI, showing that current AI systems can convincingly imitate human behavior. Interrogators often relied on linguistic style, socio-emotional factors, and knowledge-based questions to make their decisions. This research echoes Alan Turing's thought experiment, raising questions about the potential for AI to pass as human in real-world interactions. The full results are available in a paper on the arXiv preprint server.