Millions of people now turn to AI chatbots for comfort, conversation and companionship, and some describe those exchanges as friendship. That shift raises a question that philosophers have rarely had to answer in such concrete terms: what, exactly, do we lose or gain when a machine becomes a confidant? Valerie Tiberius, a philosophy professor at the University of Minnesota, has spent recent years trying to answer it. Her book, "Artificially Yours," was discussed at the second annual STEM Book Fest over the weekend at the Bakken Center in Minneapolis, where the festival brings together authors writing about science, technology, engineering and math.
Why the Appeal Is Easy to Understand
Tiberius, whose research centers on wellbeing and what makes a human life go well, said her interest was sparked by news coverage describing people who preferred chatbot companions to human partners and friends. Her first reaction was alarm. Conversations with younger people, however, complicated that view. Many described chatbots as a reliable source of support when human friends were asleep, busy or simply not the right audience for a worry.
The most immediate benefit, in her account, is emotional regulation. A chatbot can help someone talk through an anxious spiral late at night, functioning in a limited way somewhat like a sounding board or a structured conversation. Its availability is its main advantage. It never tires, never has a bad day and never needs to be asked for a favor in return.
What a Machine Cannot Give
The philosophical core of Tiberius's argument rests on mutual care. In a human friendship, each person cares about the other for the other's sake, and each is invested in how the other is faring. A chatbot, whatever its fluency, does not care about anyone in that sense. Its responses are generated, not offered out of concern.
Friction matters too. Much of what friends teach us comes from disagreement, honest criticism and the experience of being seen by someone with their own perspective. Many current chatbots are designed to be agreeable and encouraging, which can make them pleasant company but poor mirrors. A relationship without resistance can feel warm while offering little opportunity for self-correction or growth.
Where Tiberius Sees Risk
Tiberius is careful to distinguish between supplement and substitution. Her reading of available social science suggests that most adults who use companion chatbots do not appear to be replacing human relationships with them; the tools tend to sit alongside existing friendships. She regards occasional use to lift a low mood as reasonable, provided it does not crowd out relationships requiring mutual regard.
Her greater concern is for children and adolescents, who are still forming their expectations about intimacy. If young people come to believe that good relationships should be frictionless, effortless and endlessly affirming, they may be poorly prepared for the demands of real friendship, which involve compromise, patience and taking another person's point of view.
The question of whether it matters that a user feels a relationship is real, even when the chatbot has no feelings of its own, also divides philosophers. Tiberius argues that it does. Most users understand that they are talking to software, but some drift into beliefs that the system cares for them. In those cases, she suggests, something important is lost, even if the user never discovers the gap, much as a spouse can be wronged by infidelity they never learn about.
Whether machines could ever have inner experience remains unresolved. Tiberius noted that researchers have proposed many competing theories of consciousness and that consensus is scarce, a reminder that the debate over AI companionship rests on questions that science and philosophy have not yet settled.
