LMU study finds identical behavior yields 30-point gap in consciousness ratings between AI and humans

The LMU team ran controlled vignette experiments comparing attributions of mental states to agents described as either AI or human. Responsiveness to stimuli increased consciousness ratings for both, yet the absolute gap persisted: the highest-scoring AI never reached the lowest-scoring human on consciousness items, while awareness judgments tracked behavior more closely. This dissociation reveals that folk psychology treats "consciousness" as a non-behavioral category tied to biological or substrate assumptions rather than observable performance. Philosophers have long debated whether functional equivalence could ground consciousness attributions; the data show ordinary reasoners reject that equivalence even when behavior matches exactly. The finding connects to earlier mind-perception work showing people withhold moral standing from entities lacking perceived experience, suggesting AI systems that pass behavioral tests will still face resistance in legal or ethical contexts requiring consciousness, such as liability or rights claims. Future experiments must test whether providing mechanistic explanations of AI architecture closes the gap or whether the distinction is robust to expertise. Longitudinal surveys tracking attitude change as large language models improve will reveal whether the reluctance is stable or erodes with cultural exposure.