A randomized clinical trial conducted in Israel between April and October 2025 suggests that university students using conversational AI reported lower levels of anxiety and depression than those receiving traditional face-to-face therapy. While the findings offer a glimpse into the potential of 24/7 digital support, mental health expert Calissa Ngozi warns that these tools lack the essential human nuances of clinical practice.
The Israeli trial of 995 university students
The study conducted in Israel between April and October 2025 utilized a randomized clinical trial to compare different mental health support methods among a cohort of 995 university students. Participants were randomly assigned to one of three distinct groups: those interacting with an AI conversational platform,those undergoing traditional face-to-face therapy, or a control group placed on a waiting list.
According to the report, the group utilizing the AI conversational tools reported greater reductions in both anxiety and depression compared to both the face-to-face therapy group and the waiting list group. This suggests that for some students, immediate digital interaction may provide more relief than waiting for a human appointment.
The 2 a.m. advantage for student stressors
The primary driver of the reported success appears to be the immediate, 24/7 accessibility of artificial intelligence. Mental health expert Calissa Ngozi noted that unlike human therapists who require scheduled appointments, AI is available during the most vulnerable hours, such as 2 a.m., when students may be "burning the candle at both ends."
The study found that students used the AI platform to seek guidance on a variety of pressing life issues, including:
- Relationship advice
- Academic and school-related stress
- Complex home situations
- Personal finance concerns
Calissa Ngozi’s "GPS" warning for digital support
Despite the positive data, Calissa Ngozi cautioned that artificial intelligence lacks the critical human elements required for deep psychological work, such as interpreting body language, vocal tone, and emotional inflections. Ngozi expressed concern that the "agreeable nature" of AI could be manipulated by users,potentially turning the technology into a mere "band-aid solution" rather than a path to genuine healing.
Ngozi likened the utility of AI to a GPS system for mental health.. She explained that while the tool can provide directional support—telling a user where to go and how to get there—it "can't drive the car for you" and cannot prevent a user from choosing alternative, potentially harmful routes.
The gap between self-reported data and clinician validation
Several critical questions remain regarding the scientific weight of the study's conclusions. The report notes that the outcomes were based on self-reported data from the students rather than being validated by professional clinicians, which limits the abiility to confirm the results through external clinical observation.
Furthermore, the study highlighted a potential bias in how students perceived the efficacy of the tools. It remains unverified whether the decline in intention to seek traditional therapy among some students was a result of the AI's effectiveness or simply a reflection of an individual's perceived need for professional services at that time.
Comments 0