Artificial intelligence in mental healthcare: transformative potential vs. the necessity of human interaction

However, while AI offers promise for improving diagnostic precision, it has limitations, as its accuracy depends on the quality of the data it is trained on. Incomplete or biased datasets can lead to significant diagnostic errors, particularly in diverse populations, by misinterpreting symptoms or overlooking the complexity of mental health conditions. This risk is especially pronounced in mental health, where inappropriate treatments can severely impact patient wellbeing (Yan et al., 2023). For instance, a study in Ethiopia found that 39.16% of patients with severe psychiatric disorders were misdiagnosed, with rates higher among non-specialists (Ayano et al., 2021). Similarly, a Canadian study reported high misdiagnosis rates among 840 primary care patients: 65.9% for major depressive disorder, 92.7% for bipolar disorder, and over 70% for anxiety disorders (Vermani et al., 2011). Such findings underscore the inherent challenges in mental health diagnosis, which often relies on subjective doctor-patient interactions prone to inaccuracy (Yan et al., 2023). Moreover, a shortage of psychiatrists, particularly in developing countries, exacerbates the issue (Sholevar et al., 2017). In contrast, machine-based diagnoses offer several advantages, including conserving human resources, increasing efficiency, enabling large-scale assessments, and potentially reducing stigma (Uede et al., 2024); however, over-reliance on AI without adequate human oversight risks perpetuating, rather than resolving, existing issues in mental healthcare. While AI enhances diagnostics through real-time data and predictive modelling, it must be complemented by the clinical judgment of experienced professionals, as it cannot fully capture the complexity of human emotions, behaviors, and cultural factors (Graham et al., 2019;Loscalzo et al., 2017;Khare et al., 2024). Clinicians must ensure AI remains a supportive tool, not a replacement, and address risks like biased data to safeguard patient care quality (Ueda et al., 2024).The issue of accessibility in mental healthcare is a pressing concern, as many individuals, particularly in underserved or rural areas, struggle to access qualified mental health professionals (Morales et al., 2020). Despite the growing awareness of mental health issues, barriers such as high costs, long wait times, and overburdened healthcare systems make therapy inaccessible for a significant portion of the population (Kourgiantakis et al., 2023). This is where AI's role as a democratizing force becomes particularly relevant. AI-driven mental health platforms, like Woebot and Wysa, offer cost-effective alternatives to traditional therapy by providing digital interventions, particularly in cognitive-behavioral therapy (Haque Inkster et al., 2023). As Gentili (2021) highlights, technological interventions in complex systems like the human mind create 'Bio-ethical Complexity,' 'Bio-ethical Complexity' raises concerns about relying solely on AI in mental healthcare, especially as cyberattack risks grow. While robust privacy measures must be continuously updated, human error and system flaws remain significant challenges.Algorithmic bias in AI systems necessitates thorough examination, as AI models reflect the biases present in their training data, often mirroring societal inequalities across race, gender, socioeconomic status, and culture. This bias can lead to skewed diagnoses and treatment recommendations, exacerbating healthcare disparities rather than alleviating them (Celi et al., 2022). Recent cases highlight AI bias in healthcare: a U.S. hospital algorithm assigned lower risk scores to Black patients than to white patients with similar health conditions, limiting their access to care (Ledford, 2019). Another case showed a skin cancer detection model misdiagnosing darker skin tones due to predominantly white training data, reducing accuracy for non-white patients (Krakowski et al., 2024. Such examples underscore that merely refining algorithm or incorporating diverse datasets is insufficient; systemic changes in data collection, interpretation, and application are required to capture a more comprehensive and equitable view of patient needs. AI's feedback loops can entrench biases, making them harder to eliminate over time (Ferrara, 2024). Addressing this requires not just diverse datasets but also identifying implicit biases and maintaining rigorous oversight. Without these measures, AI risks deepening, rather than reducing, healthcare disparities.AI's ability to analyze extensive datasets and detect patterns that may escape human therapists offers a significant advantage, particularly in areas such as diagnostic precision and individualized care. However, AI lacks the emotional intelligence and cultural sensitivity intrinsic to human therapists, whose expertise extends beyond data to include empathy, intuition, and non-verbal communication-all critical for effective mental healthcare (Minerva Sathyan et al., 2022). Explainable AI (XAI) tools, such as SHAP (Shapley Additive exPlanations) and LIME (Local Interpretable Model-Agnostic Explanations), address these challenges by providing transparency in AI decisionmaking, allowing therapists to understand and validate AI insights without fully relinquishing control, while continuous professional development ensures that therapists use AI as a supportive tool rather than allowing it to dominate their decisions (Ali et al., 2023;Minerva & Giubilini, 2023). The future of AI-augmented therapy hinges on maintaining a balance between AI's precision and the therapist's empathy, fostering a collaborative model that enhances rather than diminishes the core relational elements of mental healthcare (Table 1). AI is a transformative force in mental healthcare, but its integration must balance technological precision with human empathy, ensuring that it complements, rather than replaces, the essential therapeutic relationship. As Topol (2019) notes, the convergence of AI and human intelligence has the potential to revolutionize healthcare by harnessing both systems' strengths. Complexity Science suggests a holistic approach, integrating dimensions like ethical, philosophical, religious, cultural and emotional dimensions with technological innovations, ensuring empathy and precision coexist in mental health treatment. Addressing adoption complexities-such as regulation, scalability, cost, and practitioner acceptance-requires robust infrastructure, phased implementations, pilot programs, and AI-human collaboration models to ensure safety, privacy, and equitable access like Wysa's approach to privacy concerns and adaptability across languages and cultures (Dinesh et al., 2014). Long-term sustainability also demands updates, ethical oversight, and resources to prevent biases and inconsistent care. While AI benefits early intervention, it may affect the therapeutic alliance, with continuous monitoring risking feelings of surveillance. Thus, AI should remain a complementary tool, carefully integrated to preserve the emotional and relational elements essential to mental healthcare. This article calls for actionable steps-such as ethical AI investment and patient-centered design-to bridge human-AI gaps in mental healthcare.