LMU Study: Humans Attribute Less Consciousness to AI Than to Humans, Even with Identical Behavior
A new LMU study reveals that people are reluctant to attribute consciousness to artificial intelligence, even when AI behaves identically to humans. A significant gap exists in how consciousness versus awareness is perceived.
Research from Ludwig-Maximilians-Universität München (LMU) indicates that humans are hesitant to ascribe consciousness to artificial intelligence (AI), even when AI exhibits human-like behavior. The study, published in the journal Cognition, highlights a distinct separation in how people perceive intelligence versus genuine consciousness in AI.
Led by Dr. Louis Longin from LMU's Chair of Philosophy of Mind, along with colleagues Dr. Bahador Bahrami and Professor Ophelia Deroy, the research directly compared how individuals attribute mental states to AI and humans under identical circumstances. This approach differed from previous studies that asked more general questions about AI's mental states.
The study involved nearly 1,100 participants who were presented with scenarios featuring either an artificial agent or a human protagonist. Participants rated the protagonist's level of consciousness and awareness of its surroundings. Results showed that increased responsiveness in behavior led to higher ratings of awareness for both AI and humans. However, when it came to 'consciousness,' a clear gap emerged: even the most responsive AI was judged as less conscious than the least responsive human.
Researchers emphasize that while people are willing to state that AI is 'aware' of its surroundings, they draw a firm line at attributing 'consciousness.' This distinction persists despite identical behavior. The choice of terminology proved significant, as more neutral terms like 'awareness' were attributed similarly to both AI and humans, suggesting that these descriptors align better with observed functions without invoking deeper concepts of subjective experience.