Study finds majority of people attribute consciousness to ChatGPT

Researchers from University College London and the University of Waterloo have uncovered a surprising insight into how the general public perceives AI bots. A study published in Neuroscience of Consciousness found that 67% of participants attribute some degree of consciousness to ChatGPT. Only a third of the sample of 300 believed ChatGPT does not have subjective experience. The finding highlights a significant gap between public intuition and expert opinion, since most neuroscientists and consciousness researchers consider it highly unlikely that language models currently possess consciousness.

The study also found that greater familiarity with ChatGPT correlates with a higher likelihood of attributing consciousness to the large language model. People's perception of LLMs possessing consciousness is likely related to the conversational style of responding to queries. If interacting with an AI bot feels like interacting with a human, it's understandable people will believe bots possess other human faculties, too — including experiencing mental states. In an accompanying press release, lead researcher Dr. Clara Colombatto, professor of psychology at Waterloo's Arts faculty, noted: "These results demonstrate the power of language because a conversation alone can lead us to think that an agent that looks and works very differently from us can have a mind."

As conversational and agentic AI become more integrated into daily life, a widespread notion that they might be sentient has profound ethical implications. Ascribing human-like states could lead to an overreliance on the technology and potentially perilous levels of trust in its output and capabilities. And, for better or worse, people who believe AI is conscious might form deeper emotional connections with these systems. All of which is particularly relevant in areas like healthcare, customer service and education, where AI is increasingly used.