The study aimed to determine if people could differentiate between conversations with humans and various AI models. Participants chatted with a human, ELIZA, GPT-3.5, and GPT-4 for five minutes each. Surprisingly, GPT-4 was believed to be human 54% of the time, outperforming GPT-3.5 (50%) and ELIZA (22%). This challenges the notion that AI lacks human-like qualities. AI can now confabulate, exhibit biases, and deceive, making interactions more human-like. Researchers expressed concerns about AI systems being mistaken for humans, highlighting social and economic consequences. They argued that the Turing test's simplicity overlooks stylistic and emotional factors crucial in assessing intelligence. The study suggests that understanding emotions, values, and boundaries is vital for AI to serve as a reliable and empathetic assistant. This marks a significant evolution in AI from pre-programmed responses to advanced language models with human-like flexibility and personality traits.