Financial Times column questions the limits of AI chatbots as therapists

In short: More people are using AI chatbots for mental health support, and a Financial Times columnist argues that the help may be limited because chatbots can only respond with words.

What's going on

A new Financial Times opinion column looks at the growing habit of using chatbots, like ChatGPT, as a kind of therapist. The writer, Jemima Kelly, describes a conversation where a friend used ideas and labels that came directly from a chatbot, such as an “anxious voice” versus an “honest voice.”

Kelly says the framework was not completely wrong and the conversation did help her feel calmer. But she argues the real benefit came from talking with a friend, not from the chatbot’s terms. Her concern is that machine-made phrases can sound certain and final, even when human feelings are complicated and unclear.

The column also points to the risk of getting too attached to labels. Words can be useful for understanding yourself, but they can also narrow your view of what is happening. Kelly suggests this can turn into a kind of “talk-only” loop, like trying to solve a problem using only a dictionary.

She cites a study from last year saying 37 percent of British adults had used an AI chatbot for mental health support. Kelly believes the number is likely higher now. She adds that some people do find the tools helpful, but a chatbot cannot replace the feeling of being with a living person.

What to watch

Expect more debate about safeguards and clear warnings for mental health chatbots, especially around when users should seek a qualified professional. As more people rely on chat-based support, the key question will be what these tools should do, and what they should not claim to do.

Source: Financial Times