As artificial intelligence systems achieve milestones that were once the exclusive domain of human cognition—drafting complex legal documents, navigating social interactions, and even demonstrating flashes of creative reasoning—a persistent philosophical and psychological question has moved from the pages of science fiction into the heart of modern debate: When a machine acts like a human, do we actually believe it is one?
According to a groundbreaking study conducted by researchers at Ludwig Maximilian University of Munich (LMU), the answer is a firm "no." Despite the increasing sophistication of large language models and autonomous agents, humans maintain a deep-seated, intuitive psychological barrier that prevents them from equating artificial intelligence with a conscious mind, regardless of how perfectly the machine mimics human behavior.
The findings, published in the journal Cognition, provide the first empirical evidence of how people categorize the mental states of machines compared to their human counterparts. The research team, led by Dr. Louis Longin of LMU’s Chair of Philosophy of Mind, alongside Dr. Bahador Bahrami and Professor Ophelia Deroy, suggests that while we may use human-centric language to describe AI, our brains continue to draw a sharp line between "intelligent performance" and "subjective experience."
The Illusion of Human-Like AI
The rise of generative AI has fundamentally altered our daily digital landscape. Reports of AI systems misleading their users, collaborating on complex projects, or appearing to pursue independent, long-term goals are now routine. When a chatbot apologizes for a mistake, expresses frustration, or adopts a specific personality, observers often find themselves tempted to use terms like "thinking," "feeling," or "deciding."
This linguistic shift has sparked concerns among ethicists and AI safety researchers: Are we being seduced by the appearance of consciousness? If a system acts as if it possesses a mind, do we eventually begin to treat it as if it truly has one? The LMU study was designed specifically to test this hypothesis by comparing human perceptions of AI against human perceptions of other humans under identical conditions.
"Whereas previous studies have typically asked general questions about whether AI actually has mental states, our study is the first to directly compare how people attribute the same mental states to AI and humans behaving in exactly the same way, under identical circumstances," explains Dr. Longin.
The researchers argue that the tendency to use mentalistic language for AI might not be a sign of true belief, but rather a reflection of a vocabulary gap. As Professor Ophelia Deroy, a philosopher and senior author on the paper, notes: "We may be using mental terms to refer to AI, but this may only be because we lack better words. After all, AI is trained to act and speak like a human, so these descriptions seem natural."
Testing the Limits of Perception
To move beyond anecdotal evidence, the LMU team conducted a series of experiments involving nearly 1,100 participants. The study structure was elegant in its simplicity: researchers presented participants with brief, controlled descriptions of agents navigating various environments.
One group was tasked with evaluating artificial agents—AI systems described as exhibiting varying levels of responsiveness. For instance, some scenarios involved an agent reacting to complex auditory input, while others involved an agent recognizing and responding to the emotional states of human interlocutors. A second group was presented with the exact same scenarios and behaviors, but the agents were described as human beings.
Participants were then asked to rate the degree to which these agents seemed "conscious" or "aware" of their surroundings. By holding the behavior constant while changing the identity of the actor, the researchers were able to isolate the specific impact that the label "AI" has on human perception.
The Disconnect Between Awareness and Consciousness
The results of the experiments revealed a nuanced psychological reality. As expected, responsiveness was a significant predictor of how participants perceived both humans and AI. When an agent—whether human or machine—reacted actively to its environment, participants were more likely to attribute higher levels of both awareness and consciousness to that agent. If a character noticed a sound or adjusted to a change in social mood, it was viewed as more "present" than an agent that remained static.
However, a profound, systemic bias emerged when the researchers looked at the specific attribute of consciousness. Even when an AI system was described as highly responsive—displaying the exact same behavioral markers as a human—it was consistently rated as less conscious than even the least responsive human in the test. The "human" label acted as a baseline for consciousness that the AI, despite its performance, could not cross.
"People are quite willing to say that an AI notices things and is aware of its surroundings," Dr. Longin observes. "But they clearly draw a line when it comes to the term ‘consciousness.’ People consistently attribute less consciousness to AI than humans—even when their behavior is identical."
Interestingly, the findings shifted when participants were asked about "awareness" rather than "consciousness." In this category, the judgments for humans and AI followed much more similar patterns. This discrepancy suggests that human cognition classifies these two terms differently. Awareness is perceived as a functional, observable capability—a matter of processing input and generating output—which people are comfortable applying to both biological and synthetic systems. Consciousness, conversely, is perceived as something deeper, touching upon the existence of an inner life, subjective qualia, and the private experience of being.
The Power of Linguistic Precision
The researchers believe these findings offer a vital lesson for how society navigates the rapid advancement of artificial intelligence. If human perception is highly sensitive to the terminology used, then the way scientists, journalists, and tech companies describe AI models has a direct impact on public understanding.
"Our study teaches us lessons for how we communicate about AI," says Dr. Bahador Bahrami, a senior author and expert in social neuroscience. "Companies and journalists often reach for hyped descriptions of AI models, and credit them with intentions, plans, or even moral conscience and hesitation."
The study suggests that such anthropomorphic language may be misleading, not because it convinces the public that machines are human, but because it creates a confused mental framework. While the public is more discriminatory than some critics might fear, the ease with which people can be nudged into attributing "awareness" to a machine means that language choice is a critical element of AI governance and ethical discourse.
The research clarifies that there is no simple, singular "mental scale" upon which humans and AI are placed. Instead, the human mind operates with a dual-track system: one that assesses external behavior and functional competence, and another that guards the concept of consciousness as a uniquely biological or human property.
As AI continues to blur the lines of behavior, this distinction between the functional and the experiential may become the most important boundary in the field of artificial intelligence. The findings from LMU indicate that as long as we maintain a clear distinction between what a machine can do and what it can feel, our perception of these systems will remain grounded in a realistic—if cautious—understanding of the technological landscape. For now, it seems the human mind remains, in its own view, a category of one.

