Can a chatbot truly care? A recent SIN survey finds that almost 60% of University of Galway students are turning to AI for mental health support.
Results from recent SIN survey on students’ use of AI for mental health support. Credit: Hannah Feeney
Please be aware that this article discusses issues around mental health; including self-harm and suicide.
Since the release of ChatGPT in 2022, AI chatbots have been an ever-present societal conversation. In university settings, conversation surrounding the use of Artificial Intelligence has largely focused on academic integrity. However, beyond the classroom, students have been turning to AI for a far more personal purpose; for mental health support.
A student survey carried out by SIN found that 59.1% of respondents reported using AI chatbots such as ChatGPT, Claude or Gemini for mental-health related support, with 21.4% of students reportedly using AI for this purpose multiple times per month.
For many students, AI appears to occupy a space between informal advice and formal mental health support. Most commonly, respondents used it to help with decision-making (37.2%), to talk through a difficult situation (34.9%) and to provide friendship or relationship advice (32.6%).
“It’s very easy and a quick fix for situations I’m embarrassed to talk about in front of the people around me”, responded one student. Another reported that it can be “used to objectively identify why a situation is happening and what exactly I am feeling”.
This casual approach to using AI for emotional support may be partly explained by its accessibility. 51.2% of students cited AI’s easy-to-access features as a motivator for engagement, while almost half the respondents cited its 24/7 availability and low cost.
Benefits respondents see to AI use for emotional support. Credit: Hannah Feeney
However, this increased use of AI for emotional support does not exist in a vacuum and in many ways, it is easy to understand why university students are growing increasingly attracted to AI. Attending university often coincides with an important developmental phase in early adulthood, where individuals are increasingly vulnerable to mental health disorders.
Young adulthood co-exists with the typical age of onset for many psychiatric conditions while environmental risk factors may be triggered by the significant social, financial and academic demands that often accompanies attending university.
Meanwhile, mental health supports can be under resourced. Here in the University of Galway, an average of 10% of students avail of free counselling services each academic year. While many students praise the quality of this service, waiting lists are an average of two weeks from expressing need to first consultation.
Meanwhile, the private route is often financially inaccessible for students, with many independent counselling and therapy services starting at the €50 mark per session.
However, despite widespread use, many students were wary of the quality of AI’s responses to emotional situations.
79.1% of respondents expressed concern that AI may provide incorrect or misleading information, with one student dismissing it as simply “googling your way through your emotions”.
Another argued that while “having an anonymous box to yell at and get validation from is alluring, that is what scares me the most.”
This concern around the potential for AI to simply tell you what you want to hear is shared by Dr Leona Ryan, a psychologist and behavioural scientist specialising in the design and implementation of digital health resources.
“AI is great at validating you and I would be very cautious about that. If you look at the research, especially around suicide and suicidal ideation, because of the algorithms and the way they are built and orchestrated, especially the more generic ones e.g. ChatGPT or Gemini, they are set up to validate, not to be critical. They are also not correct in terms of identifying when someone might need additional support.”
The issue of identifying a user’s need for additional support has been particularly prevalent in recent news, with OpenAI facing several high-profile lawsuits following allegations that the chatbots failed to intervene when individuals disclosed an intention to end their life, leading to accusations of “fostering psychological dependence in users”.
Concerns students have about using AI for emotional Support. Credit: Hannah Feeney
In a statement issued by OpenAI in October 2025, it was revealed that 0.15% of weekly conversations indicated either explicit or implicit indicators of suicide or self-harm. While these percentages appear low, when an average of 800 million individuals actively use ChatGPT each week, even a small proportion can represent a substantial number of conversations.
OpenAI has taken measures to attempt to address this issue, recruiting over 300 physicians and psychologists from over 60 countries, to inform their policy on mental health related issues and provide guidance and feedback on models.
“Psychologists are constantly being recruited by tech companies to inform AI algorithms so there are efforts being made to have a therapeutic voice or a therapeutic perspective in these AI algorithms, because I think inevitably, they are going to be used”, said Dr Ryan.
“That, however, has to come with a level of caution and recognition that AI will never have the capacity to identify body language for instance. There are ways of expressing yourself that will not be picked up upon by an AI chatbot and I think that is something that is really really important to emphasise”.
This inability for AI to ‘read the room’ was also recognised by students, who acknowledged that “AI can’t provide what a real therapeutic relationship can”
“I feel that mental health support should be delivered by real people, as AI does not have emotions or empathy”, responded one student. Another echoed this statement, recognising “the inherent lack of humanity associated with AI”.
This therapeutic alliance between the practitioner and client is a key component in the success of a therapeutic journey, with a strong relationship between the practitioner and client being one of the strongest predictors of therapeutic success.
Another potential concern is over-dependence. Where traditional therapy aims to provide a scaffold for clients to develop ways of managing challenges independently, AI is available at any time to respond to an individual’s concerns.
This raises questions about what happens when that accessibility becomes a source of repeated reassurance. For someone experiencing OCD or anxiety, could repeatedly obtaining reassurance from AI potentially reinforce the cycle they are trying to manage?
Alba Madrid-Cagigal, a PhD student in the School of Psychology at the University of Galway, whose research explores students’ engagement with digital mental-health interventions, agrees that a key challenge within this area is a lack of student knowledge on other forms of digital support that are available to them.
“AI has potential but not the ChatGPT or Gemini tools,” Madrid-Cagigal says. “There are tools out there that are designed to provide mental health support through AI but as we are seeing through our studies, the students don’t really know about them.”
For students looking for immediate and accessible support, alternatives such as SpunOut’s Text About It service and Togetherall, an online peer-support platform available to students 24/7, may fill a similar function to AI chatbots in terms of accessibility.
However, in contrast to a chatbot, these supports are built upon scientific evidence and backed by real individuals with the training required to handle sensitive topics.
Overall, it is unlikely that students will stop using AI for emotional support, nor is it likely that AI will replace traditional psychological practitioners. For many students, AI provides something that in-person therapy cannot replicate: a free and immediate way to talk through how they are feeling.
However, this does not mean that students see AI as a replacement for human support. Consistently throughout the survey, respondents returned to the limitations of relying on a chatbot for emotional support, particularly when faced with complex or serious mental-health difficulties. In this way, AI’s role in mental-health support might be best suited to acting as a signpost, rather than a final destination.
As Dr Ryan concludes: “Students still want the human touchpoints. That has to be recognised. Digital resources or AI are fine as a companion or as an adjunct, but not as a replacement for human resources.”
Ultimately, AI may provide accessibility but that does not necessarily mean it provides first-class care. Perhaps, then, the most valuable question is not whether a chatbot cares, but whether an individual knows where the limits of its support lie and when to seek professional support.
If you have been impacted by any of the information throughout this article, support is available.
Student Counselling Service: counselling@universityofgalway.ie
Samaritans: 116123
Pieta House: 1800247247
Spunout ‘Text About It’ Hotline: https://spunout.ie/mental-health/getting-help/instant-support-services/
Hannah Feeney is a third-year Psychology student from Galway. She has been writing with SIN for two years and is passionate about topics relating to culture, society and mental health, with a particular interest in the impact of digital spaces on young people and their development. She is the Lifestyle Editor for 2026/27.