The Emancipatory Role of Technology in Mental Healthcare: The Case of Conversational Artificial Intelligence and Digital Platforms

Abstract

The mental health crisis has been rapidly growing over the last few decades and was further exacerbated by the COVID-19 pandemic, turning it into a deadly global challenge that leads to the loss of 12 billion working days each year, accounts for more than 14% of all deaths, and will cost the world's economy $16 trillion by 2030. Half of the world's population will experience a mental health disorder during their lifetime. Accessibility, affordability, lack of trained personnel, and stigmatization are among the most pressing mental health issues of our time. They act as serious obstacles to comprehensive solutions for this critical global problem, a situation that is getting worse at an alarming rate given the widening gap between the supply (i.e., readily available solutions) and demand (i.e., global need for mental health services). Conversational artificial intelligence (e.g., chatbots and large language models) and digital telehealth platforms, given their unique features (e.g., low operational costs, and scalability) have what is required to address the above-mentioned obstacles and thus be regarded as practical solutions for this grand challenge of our time. In a series of theoretical, empirical, and experimental studies (nine full studies, eight pilot studies, and two supplementary empirical exercises) and using various quantitative and qualitative methods (ranging from grounded theory to unsupervised machine learning), we examine different capabilities of these technological artifacts in the context of mental healthcare and uncover previously unknown, occasionally counterintuitive, and highly important insights about these promising solutions. We also propose a novel investigation of the best approaches for empathetic AI agents based on the latest generations of generative chatbots like large language models. Our results expand prior theory, challenge previous assumptions, and inform scholars and practitioners about the IT system features that encourage people to disclose risky and stigmatized information, characteristics of chatbots and LLMs that can increase the use of these AI agents, maximize patients' information self-disclosure to them, enhance patients' willingness to follow their advice, while improving their engagement and satisfaction with these agents.