When Care Becomes a Crutch: Why We Need to Rethink Support and Treatment

## Introduction We are in the middle of a technological revolution that also reaches the most intimate parts of our lives: how we understand and handle mental health. Chatbots that promise conversational support, algorithms that tailor exercises, and generative AI that imitates therapeutic empathy — all of this promises greater accessibility to help. At the same time, experts warn that well-intentioned practices at home and in communities may leave children and young people less prepared for life's challenges. According to the VG article 'Experts: The parenting style that makes children less able to cope', a parental approach that some view as perfectly normal can in reality harm children more deeply than expected. This is not a contradiction we can ignore: technological solutions for mental health meet generations that may not have developed basic coping skills. In this opinion piece I combine insights from research and debate on parenting with ongoing discussions about artificial intelligence in therapy. The aim is not to dismiss the technology, but to point out what is at stake if we roll out AI therapists without considering human and social contexts. The article is written for professionals, parents, decision-makers and anyone interested in how we best support mental health in a digital age. ## Why parenting style still frames mental health As reported in the VG article 'Experts: The parenting style that makes children less able to cope', experts point out that certain features of parenting can have major consequences for children’s ability to handle challenges. The article describes findings showing that an apparently normal way of caring can be harmful — especially when it includes overprotection or a lack of support for emotional independence. This matters because the development of coping skills, resilience and emotional regulation happens in close interaction with parents or primary caregivers. When parents continuously solve problems for the child, they reduce the child's chance to experience and learn from difficulties. If a child never gets to practise tolerating frustration, managing discomfort, or finding their own solutions, they become vulnerable later in life — at school, at work, and in relationships. The VG article highlights that 'experts point to two particularly worrying traits', and this reflects concern in the field: well-meaning care can sometimes hinder the development of autonomy and robustness. This is a fundamental point that few technological forms can compensate for if the social learning never takes place. For those who work with or plan digital health services, this is necessary background: therapy technology meets individuals shaped by relationships. If we do not account for how parental and environmental factors have affected users' skills in handling life's challenges, we risk offering tools that act as temporary band-aids rather than lasting help. ## What artificial intelligence in therapy can actually offer The term AI therapist covers a range of tools: from chatbots offering exercises inspired by cognitive behavioral therapy to more advanced models that can generate therapeutic dialogues and tailored interventions. The benefits are real. Generative AI mental-health solutions can provide 24/7 availability, anonymity for those who fear stigma, and a lower threshold for seeking help. For many, a chatbot psychologist can be the first step toward articulating problems and learning basic strategies for emotional regulation. Furthermore, digital tools can support early detection: patterns in language, sleep, or activity levels can be indicators that help professionals identify risk early. When such systems are used as support tools in combination with human treatment — not as replacements — they can relieve pressure on health services and expand the reach of available services. Still, it is crucial to understand the limits. AI cannot 'feel' or understand human experience in the same way a person can. Empathy is not just appropriate word choice; it consists of holistic understanding, body language, long-term continuity and the ability to hold complex, contextual information over time. Generative AI can mimic these features, but without genuine intention or moral responsibility. Therefore solutions must be designed with clear boundaries and good integration with human follow-up. ## Risks and ethical dilemmas of AI therapy When we move from potential to implementation, a complex set of ethical and safety challenges emerges. Privacy is obvious: conversations about mental health contain extremely sensitive information, and how these data are stored, who has access, and how they are used determines whether trust in digital services is maintained. In addition there is the risk of mistreatment. An AI model can give incorrect or unsuitable advice, misunderstand cultural context, or reinforce harmful narratives it has learned from biased data. The question of responsibility is also critical: who is accountable when a chatbot provides harmful guidance — the developer, the provider, health authorities, or the user? Lack of clear regulation can lead to situations where users receive poor or dangerous help without manageable complaint mechanisms. We must also be aware of how algorithms can maintain or amplify inequality. If training data largely reflect certain groups, AI therapists may be less accurate for minorities or non-Western cultures. Finally, there is a psychological risk: excessive dependence on technology for emotional support. If generations grow up without learning to handle adversity — as the experts in VG fear in connection with certain parenting styles — we risk a population that increasingly seeks immediate digital consolation rather than developing durable coping strategies. Thus AI can become a short-term solution that cements weaker skills over time. ## The connection between parenting style and use of AI therapy This is where the discussion from the VG article becomes especially relevant to the development of digital health solutions. If we accept that some parents today, with the best intentions, reduce children’s opportunities to learn from adversity, we must also consider how this affects the need for and effectiveness of AI therapy. A child who has not learned emotional tolerance may more readily turn to digital tools to avoid discomfort. This creates a double challenge: the user needs help with basic skills, while overly frequent use of an AI therapist without human follow-up can delay the development of those skills. According to VG there are particularly two worrying traits of modern parenting; this insight should inform how AI solutions are designed. First, AI therapists should not act as problem-solvers that remove all friction from the user's life. Instead they should be designed to facilitate gradual exposure, guided problem-solving and strengthening of autonomy. Second, systems must be transparent about their limitations: users and parents need to know that a chatbot is not a substitute for the development of relational skills or for human therapy when required. Seeing AI as a supplement, not a substitute, is key. This implies concrete product choices: built-in exercises that train tolerance for discomfort, routines that encourage social interaction and real human support, and mechanisms that flag the need for clinical follow-up in time. ## Practical recommendations: Safe implementation and responsible use How do we move from concern to action? Here are concrete principles and measures that should underlie responsible rollout of AI therapy: 1. Human-in-the-loop: AI should support clinicians, not replace them. Systems must have clear procedures for follow-up when a user shows severe symptoms. 2. Transparent communication: All users must be informed about what AI can and cannot do. This includes limitations, training data and potential biases. 3. Privacy and data minimization: Conversations should be encrypted, storage limited, and users must control who accesses their data. Ethical guidelines must be set for secondary use of anonymized data. 4. Evaluability and clinical trials: Before commercial rollout, digital therapists must undergo controlled studies that evaluate effectiveness, risk and for whom the tools work best. 5. Cultural and developmental adaptation: Systems must be trained and tested on diverse populations to avoid bias. 6. Parenting guides and education: Parallel to technological offerings, society should invest in training for parents — on how to promote children's coping skills and when professional help is needed. This directly addresses the issues raised by VG about parenting's impact on children's capacity to handle life. 7. Regulation and accountability: Authorities must set clear frameworks for responsibility, safety and approval of digital therapists on par with medical devices. Implementing these principles requires collaboration between technologists, clinicians, users and policymakers. To succeed, not only must the technology be good, but the infrastructure around it as well: access to human support, healthy digital habits and a societal understanding of what good parenting entails. ## A balanced future: Innovation with a human touch As a society we face a crossroads. We can either let technological solutions fill a need without demanding how they are shaped, or we can actively steer development so that AI therapists become tools that promote lasting life skills. According to the VG article, there is cause for concern when a parenting style 'that some see as perfectly normal' actually undermines children's ability to cope with life. This must be part of the premise when we design and regulate generative AI mental-health technology. My clear appeal is that we do not isolate technology from the social factors that shape users. AI therapists should incorporate educational elements that strengthen autonomy, and systems must always be anchored in good, human follow-up services. Investment in parent education and community-based coping programs is as important as investment in technological innovation. If we succeed, we can get the best of both worlds: the accessibility and scalability of AI combined with human accountability and a long-term focus on building resilient individuals. If we fail, we risk that digital care becomes a crutch that conceals — but does not solve — fundamental problems. ## Conclusion Mental-health technology is a powerful opportunity — but also a risk. As reported in the VG article about parenting styles that weaken children's ability to cope, there are fundamental relational and learning challenges that a chatbot alone cannot solve. AI therapy can be a useful supplement, particularly for increasing access and early detection, but the ethical, legal and psychological implications must be handled seriously. We need stricter regulation, clearer lines of responsibility, transparent solutions and, most importantly, an integrated approach that builds human competence — not replaces it. Parents, clinicians, developers and policymakers must collaborate to ensure the next generation receives both technological access and human strength. Only then can we ensure that care and treatment lead to genuine mastery, not permanent dependence. ## Sources - VG: 'Experts: The parenting style that makes children less able to cope', published 12.04.25, updated 13.04.25. URL: https://www.vg.no/forbruker/i/4B8bJ9/eksperter-foreldrestilen-som-gjoer-at-barn-takler-livet-daarligere