Blog post Part of series: Artificial Intelligence in educational research and practice
My AI companion? Rethinking support, trust and wellbeing in education
A school pupil struggling with homework, or a university student facing a difficult assignment may no longer turn first to a teacher, parent or friend. Instead, they may open an artificial intelligence (AI) chatbot. What begins as help with homework can quickly become something more personal: reassurance after a poor grade, encouragement before an exam, feedback on ideas, or simply the feeling of being listened to. My work as an education mental health practitioner and public health researcher has led me to consider what happens when a learning tool starts to feel like a source of care. This blog post considers what that shift means for learners’ trust, help-seeking and wellbeing when human support is scarce.
Debates about artificial intelligence in education often focus on . These concerns matter, but they do not capture the full picture of how young people are beginning to use AI. A found that one-third of teenage AI-companion users chose AI over a person for serious conversations. A more pressing issue may be learners’ growing emotional dependency on these systems. If we only frame AI as a tool for cheating or efficiency, we risk missing how it is quietly changing the way some learners seek support.
Generative AI can support learning. It can explain ideas quickly, adapt language to a learner’s level, support revision and reduce anxiety about asking questions. For learners who lack confidence or are learning in an additional language, generative AI may provide a flexible space for practice, repeated explanation and personalised feedback. A suggests possible benefits for learner confidence, motivation and engagement, while a reports benefits for language proficiency and related learning outcomes, although effects vary across learners and settings. Used well, it can help students engage with difficult material in a less intimidating way.
But the idea of AI as a ‘companion’ needs careful attention. A companion suggests trust, presence and concern. , offering reassurance and a space to reflect, even though it cannot genuinely understand or empathise. However convincing its responses may seem, a chatbot does not share an educator’s safeguarding responsibilities or duty of care, and may not reliably recognise indirect signs of distress (Dvir et al., 2025). If a learner trusts a chatbot more than a teacher, that is not just a technological issue. It is also raises concerns about care, safety and relationships.
‘The key question is not simply whether students use AI, but what happens when they begin to depend on it for reassurance or emotional support.’
This may matter particularly for vulnerable learners. Learners experiencing isolation, anxiety, disadvantage or exclusion may be especially drawn to AI because it is available, patient and seemingly non-judgmental. That makes its emotional appeal important to understand. Prinsloo et al. (2024) highlight the challenges AI-powered systems create for digital wellbeing, particularly among vulnerable students. The key question is not simply whether students use AI, but what happens when they begin to depend on it for reassurance or emotional support.
Schools and universities, therefore, need to develop both AI literacy and wellbeing literacy. Students should learn how to judge AI-generated information, recognise bias, protect privacy and understand the limits of automated feedback. After a , university students reported greater confidence, better understanding of institutional policy and clearer ideas about using generative AI appropriately in their learning. They should also be encouraged to think about why they are using AI. Is it helping them learn, or replacing human support? Is it helping them understand an idea, or stopping them from speaking to a teacher, peer, parent, counsellor or trusted adult?
This is not an argument against AI in education. Used carefully, it may widen access and make asking questions less intimidating. Yet AI companions can . They should complement, not replace, human support. AI literacy should help learners distinguish simulated care from human support, recognise when further help is needed and know where to find trusted support.
黑料网911 has recognised that AI is becoming increasingly embedded in educational research and practice, influencing pedagogy, curriculum, assessment and ethics. That makes this a timely issue for educators, researchers and policymakers alike. If AI becomes more conversational and emotionally persuasive, students may start to experience these systems not only as sources of information but as trusted companions.
As conversational AI becomes more socially responsive, attachment may matter as much as assessment. Some learners may , despite the relationship being simulated. The challenge is to ensure that AI supports learning without displacing the human relationships that provide care, trust, safeguarding and a sense of belonging.
References
Dvir, Y., Moore, P., & Kelly, M. (2025). AI chatbots and youth mental health: Practical recommendations for clinicians. Journal of the American Academy of Child & Adolescent Psychiatry. Advance online publication.
Prinsloo, P., Khalil, M., & Slade, S. (2024). Vulnerable student digital wellbeing in AI-powered educational decision support systems in higher education. British Journal of Educational Technology, 55(5), 2075–2092.
Artificial intelligence declaration: The author used Microsoft Copilot to assist with proofreading this blog post. The author takes full responsibility for the accuracy and originality of this publication and complied with the terms of .