In the middle of the night, if your heart suddenly starts beating a little faster, what would you do? In the past, many people would first search on Google, but now some people directly turn to ChatGPT: “What’s happening to me?”
AI is indeed convenient. No need to make an appointment, it doesn’t mind your chatter, and it’s available even at two in the morning. However, a study involving nearly 100,000 American college students found that those who used ChatGPT, Claude, and other generative AI to ask about health issues had a higher proportion of positive screenings for anxiety and depression.
Whether it is AI that makes people more anxious by asking more questions, or if anxious individuals are more inclined to seek answers from AI, is still unclear.
The study was published online in the Journal of Counseling & Development on September 16. Researchers analyzed data from 96,205 American college students and found that among those who used AI to ask about health issues, about 52% screened positive for clinically significant levels of anxiety, compared to around 43% of those who did not use AI; the proportions of positive depression screenings were about 47% and 38% respectively.
“Positive screening” here does not mean a hospital diagnosis. After considering factors such as age, gender, race and ethnicity, socioeconomic status, etc., researchers found that the odds of positive anxiety screenings were 52% higher for those who used AI to ask about health issues, and the odds of positive depression screenings were 46% higher.
However, this only indicates a correlation between the two, not necessarily that “AI is causing people to be more anxious”.
This scenario is easy to imagine. For instance, experiencing sudden headaches and asking AI: “Could there be a problem?” AI lists out several reasons, and one of them seems to resonate, prompting the user to ask further: “What if I also feel a bit nauseous?” and then recalling feeling dizzy yesterday.
What was initially a single question has now spawned a second and third question.
Yusen Zhai, Assistant Professor at the College of Education at the University of Florida, and one of the study authors, mentioned that repeatedly asking AI for reassurance may reinforce rumination and health anxiety.
This is where conversational AI stands out. While a search engine provides you with a list of links, AI can keep engaging in conversation with you. For those who are already concerned about their bodies, seeking reassurance from AI initially might end up raising more questions as more answers are uncovered.
However, things could also be entirely reversed: those who are already more anxious or attentive to bodily changes may naturally be more inclined to turn to AI for health inquiries.
So, whether “people who ask AI are more anxious” or “anxious individuals are more likely to ask AI” remains unanswered by this study. It only reveals a clear association between the two among this group of American college students.
Researchers also do not advocate against using AI to seek health information. For some students, AI might be the first step in seeking help. It’s just that AI’s responses may be incorrect and lack understanding of a person’s complete medical history and actual physical condition.