According to Experts.news, regular interaction with artificial intelligence and social robots may gradually influence how people communicate, perceive themselves, and shape their behavior, according to a study by researchers from the United Kingdom, Denmark, and Sweden published on August 19, 2026, in the journal AI & Society.
The authors of the study, from the University of Birmingham, Aarhus University, and Linnaeus University, coined the term “robotoid humanness.” By this term, the researchers mean a possible process in which, through repeated interaction with AI, a person begins to adapt their behavior to the machine’s logic—becoming more predictable, standardized, and amenable to algorithmic processing.
This effect could be particularly noticeable in the service sector, where AI is already used in retail, the hospitality industry, tourism, and healthcare. Modern digital assistants and robots mimic human speech and emotional cues and personalize their responses to build trust and facilitate interaction with customers.
Researchers suggest that this process may be two-way. Machines are becoming increasingly human-like, but humans may also unconsciously adopt communication patterns from their interactions with machines.
“The consumer acts, the robot responds, and through repeated interaction, the person eventually adopts this communication model,” explains Selchen Ozturkjan of Linnaeus University, one of the study’s authors. According to her, people’s natural tendency to mimic the behavior of their conversation partner can lead users to begin replicating the communicative traits of AI.
The authors describe a three-stage mechanism. First, a person begins to perceive their interaction with the AI as a social one. Then, the algorithm creates a simplified digital model of the user and feeds it back to the user in the form of personalized responses. Subsequently, the user may gradually adapt their self-presentation and behavior to the patterns that the system more easily recognizes and encourages.
The researchers identify one of the risks as the formation of a kind of feedback loop, in which the algorithm constantly reinforces the user’s characteristics that it already knows. Unlike communication with people, where a conversation partner might unexpectedly disagree, misunderstand, or offer a completely different perspective, algorithmic systems are typically designed for consistency, convenience, and personalization. This has the potential to reinforce existing views and reduce the diversity of self-expression.
At the same time, the authors emphasize an important limitation of their work: this is a theoretical study, not an experiment involving human participants. Scientists have not yet proven that prolonged interaction with ChatGPT, robots, or other artificial intelligence systems actually makes people more “robot-like.” The proposed concept must be tested in further empirical studies.
The research paper “Robotoid humanness: when selfhood becomes machine-legible” was published on August 19, 2026, in the journal AI & Society. The ScienceAlert article highlighting the study was published on August 31.
Research article in AI & Society
