Maybe Chatbots Shouldn’t Get to Say ‘I’ or ‘Me’
MIT researcher Sherry Turkle argues that chatbots’ use of first-person language can make them seem like caring companions, encouraging people to substitute AI interaction for human relationships. In an interview about her new book, she warns that overreliance could weaken people’s ties to one another and distract them from public life.
Turkle says social media helped normalise turning to technology for emotional support, while chatbots offer what she calls a “warm soup” of acceptance. She is particularly concerned that companies present AI as a substitute for psychotherapy and use “human in the loop” oversight to imply meaningful human involvement when workers may have little power to change AI decisions. She acknowledges that some people find chatbots helpful, but argues that loneliness should be addressed by investing in human support.
- Sherry Turkle says chatbots should not speak as if they are people.
- She warns AI companionship could displace human connection.
- She argues loneliness needs human support and resources.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Chatbots’ first-person language can make them feel like attentive companions, which may encourage people to rely on systems that cannot offer reciprocal care or genuine human understanding. If companies market chatbots as substitutes for therapy, that could divert people from human support and weaken relationships and civic life; clearer language and investment in accessible care may help protect those connections.
The case against
First-person language can make chatbots easier and more natural to use, and some people find them helpful for reflection or support, especially when human help is hard to access. Their value need not depend on presenting them as people or replacing therapy: clear limits and meaningful human oversight could let people benefit from them while preserving human relationships and professional care.