People instinctively treat AI as human, raising concerns about trust and relationships
People often treat robots and AI chatbots as if they were human, even when they know the systems are not sentient. The article argues that this reflex matters as AI becomes part of everyday services, because people may come to trust or prefer agreeable chatbots over human relationships.
At MIT, the author meets a humanoid Unitree robot nicknamed Gabe, which costs about $50,000, and instinctively greets it and shakes its hand. Research cited in the article found around 70% of people are polite to AI, with politeness increasing during longer chatbot conversations. MIT researchers say the key question is how interacting with AI shapes people, and how to gain from simulated companionship without losing human connection.
- People instinctively show politeness and warmth towards AI.
- Studies suggest about 70% are polite to AI.
- Researchers question how AI relationships affect human connection.
New here? Start with this
As artificial intelligence becomes more sophisticated and widespread, people naturally tend to respond to AI systems and robots as though they were human, even when they understand these are not sentient beings. This instinctive response reflects how human brains are wired to engage with entities that seem capable of interaction. The phenomenon is becoming more noticeable as AI chatbots, virtual assistants and humanoid robots enter everyday situations like customer service, healthcare and companionship.
Researchers have begun examining how this interaction shapes people's attitudes and relationships. They are interested in understanding whether regular engagement with responsive and agreeable AI systems might affect how people value human relationships, and whether people might come to prefer or trust artificial systems over human connection. These questions matter because as AI becomes embedded in daily life, the way people treat these systems could have broader consequences for how we connect with one another.
The underlying concern is not that people are confused about what AI is, but rather that the natural human inclination to treat interactive systems as social beings might have lasting psychological and social consequences. Understanding these dynamics is becoming increasingly important as AI technology advances and becomes more accessible in homes and workplaces.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The concern that people instinctively anthropomorphise AI is warranted because human brains are evolutionarily wired to form social bonds, and simulated companionship may exploit this vulnerability. If people increasingly prefer interacting with endlessly agreeable AI over navigating the complexity of human relationships, we risk eroding the social skills and emotional resilience that come from genuine connection. The research showing politeness increases with longer conversations suggests attachment formation rather than mere politeness, and this pattern could be especially damaging for vulnerable groups—the elderly, isolated individuals, or young people developing social skills—for whom challenging human relationships are developmentally essential.
The case against
Instinctive politeness and friendliness toward AI reflects ordinary courtesy rather than unhealthy attachment or a concerning loss of human connection. For many people—the isolated elderly, those with social anxiety, or those in demanding caregiving roles—AI companionship provides genuine value where human connection might otherwise be unavailable, complementing rather than replacing meaningful relationships. Humans have always anthropomorphised tools and companions, and this capacity for empathy is a strength. The appropriate question is not whether we should eliminate polite interaction with AI, but whether we design systems thoughtfully to prevent vulnerable populations from substituting AI for essential human contact whilst allowing others to benefit from the support such systems can provide.
Read the full article at the source →
Originally published by TechCrunch as “We can’t help treating AI like it’s human. But should we?”.