CAI Currents is a series exploring the latest news on tech and mental health from a psychoanalytic perspective. It is published by TAP in partnership with the American Psychoanalytic Association President’s Commission on Artificial Intelligence.
CAI Currents, May 26, 2026
Thanks for Nothing? People Follow Chatbot Advice Whether It Helps or Not by Molly Pachan
Is Your Therapist Ready to Talk About Your Chatbot? by Falisha Oser
Thanks for Nothing? People Follow Chatbot Advice Whether It Helps or Not
Molly Pachan, PhD
As the saying goes, advice is cheap. Turns out, so is following it, whether it’s coming from a college roommate or a chatbot.
A recent study documents this, at least for chatbot advice. A Human-Computer Interaction article reported that 79% of participants followed chatbot advice about health, careers, or relationships after just one conversation. However, this digital guidance did not improve their well-being at follow-up three weeks later when compared to a control group that discussed hobbies and interests.
Most users found guidance they receive from the “answer machine” pretty compelling for mundane as well as serious topics. The study’s authors found that participants were equally likely to follow advice when the stakes were high and the problem was severe. The authors concluded that AI chatbots are “highly influential” advisors in a short burst of conversation.
While the rates of inappropriate or unsafe advice in this study were low, the authors note that with the scale of AI adoption those rates translate to large numbers of people. Of course, because research lags behind technological developments, the study used AI models no longer on the frontier. Plus, the sample did not focus on clinical risk, users with psychiatric conditions, or people with temporary or temperamentally low ego strength.
Nevertheless, that people follow LLM advice even when it does not help them raises troubling questions about whether or not we are building humane technologies, or technologies misaligned with human needs. Are guardrails being developed with a power commensurate with the power of emerging AI to push engagement, products, and politics not aligned with well-being? And how can a psychoanalytic focus on the value of an examined life help people humanize a potentially dehumanizing technology?
Is Your Therapist Ready to Talk About Your Chatbot?
Falisha Oser, LMHC-D
When people have a sensitive or vulnerable issue to discuss, research suggests more and more of them are talking to a chatbot. They’re choosing AI over friends, family, and no surprise, mental health professionals.
The resulting consulting room silence is the topic of a recently published JAMA Psychiatry commentary titled “Patients Use AI—Clinicians Should Ask How.” They argue that AI use is like any other help-seeking behavior and should be routinely assessed. As they say, “understanding how patients cope is essential to helping them.” They go on to argue that therapists should reinforce what helps, correct AI’s sycophantic or misleading responses, and help patients develop a critical eye towards chatbot interactions.
They also claim therapists should gain a basic level of AI literacy as a way to help patients navigate the terrain of AI relations. One implication not stated in the article is that people should start asking their mental health professionals whether they have the AI literacy they need to be helpful to them.
Learn more: Get news on other CAI projects.
Write to TAP’s advice column, Ask a Psychoanalyst.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.