Researchers: Frequent AI Conversations May Teach Humans Machine-Like Behavior

An international team of researchers has described a phenomenon called "robotoid humanness" in the journal AI & Society, a theory suggesting that repeated interactions with AI systems may push people toward more predictable, ritualized patterns of behavior.

What the theory describes

The research team analyzed how features of today's AI systems, such as memory of past conversations, adaptive learning, personalized communication and simulated emotional engagement, affect how people perceive themselves. At the center of the model is a feedback loop: the user adjusts their language and behavior to the robot or chatbot, the system learns from these signals and becomes even more human-like in response, which in turn deepens the adjustment on the human side.

The researchers write that interacting with robots that display human-like behaviors, such as adaptive learning, personalized communication and empathetic engagement, can reinforce, transform or destabilize a consumer's self-perception. This wording suggests the effect is not necessarily negative, but depends on context and the frequency of contact with the technology.

The mirroring mechanism

Machine learning and AI can amplify this process by adapting the robot's behavior based on user input and enabling more personalized, more human-like mimicry on the robot's part - Dr. Selcen Ozturkcan, Linnaeus University

The authors stress that the mimicry-mirroring loop works in both directions. The more data a system collects about a user's preferences and communication style, the more accurately it mimics human behavior, which in turn prompts the person to further align themselves with patterns the machine understands best: simpler, more repetitive, easier to predict.

The analytical framework proposed by the researchers includes four elements: the service environment in which the interaction takes place, the consumer's expectations of the system, the appearance and communication style of the robot or chatbot, and the commitments made by both parties during the conversation. According to the authors, each of these elements modulates the strength of the "robotoid humanness" effect.

Risks and caveats

The researchers list several areas of risk they say require further study. These include dependence on validation of one's self-worth through the AI system's responses, reinforcement of confirmation bias due to repeated exposure to algorithmic feedback, and the influence of cultural differences, personality traits and a user's technology readiness on the strength of the described phenomenon.

A key caveat concerns methodology. The paper is purely theoretical and is based on a synthesis of prior research on human-robot interaction rather than a new experiment involving people tracked over time. The authors themselves call for "careful consideration of the cognitive and ethical implications of shaping consumer identity," suggesting they view their model as a starting point for further empirical research rather than definitive proof.

Why it matters beyond the lab

The theory arrives at a moment when more and more everyday interactions, from customer service to voice assistants to tools like ChatGPT or Claude, take the form of a conversation with a system that remembers context and adjusts its tone to the person it's talking to. If the mechanism described by the team from Birmingham, Aarhus and Linnaeus actually holds up in practice, it would mean that the design of conversational interfaces itself, not just the content of AI responses, influences how people communicate with one another.

For companies deploying customer service chatbots or AI assistants, the paper's conclusions suggest an added dimension of design responsibility that goes beyond the accuracy or effectiveness of responses alone. The authors do not propose specific regulations or guidelines, but they point to the direction future empirical research should take, including long-term measurements among real users who interact intensively with conversational systems.