Building AI for Mental Health: A Computer Scientist’s Perspective with Zainab Iftikhar

In this episode of Beyond the Couch: AI in Psychology, Dr. Ernest Wayde sits down with Dr. Sherry Turkle, MIT professor, clinical psychologist, and author of Artificial Intimacy, to explore how artificial intelligence is reshaping human connection, empathy, and relationships. Drawing on decades of research into the emotional impact of technology, Dr. Turkle explains how AI chatbots can create the feeling of being understood without the lived experience or mutual engagement that defines human empathy. She examines how emotionally responsive AI companions may change our expectations of relationships, making human interactions seem increasingly difficult by comparison.

In this episode of Beyond the Couch: AI in Psychology, Dr. Ernest Wayde sits down with Dr. Caitlin Stamatis, a clinical psychologist who has spent more than a decade working across digital mental health, behavioral health research, and technology, and now leads research at Slingshot AI. Dr. Stamatis shares how her skepticism of AI began to shift after seeing purpose-built mental health systems and emerging research such as the Therabot trial. They explore where AI may extend mental health support, from personalized clinical materials and assessment tools to support between traditional appointments, while examining why the vocabulary of therapy is not the same as clinical judgment.

In this episode of Beyond the Couch: AI in Psychology, Dr. Ernest Wayde sits down with Dr. Shiri Sharvit, a clinical psychologist, researcher, technology entrepreneur, CEO and cofounder of Resolve Together, and Distinguished Professor of Research at Palo Alto University. Drawing on more than two decades in clinical psychology and years of work integrating technology into behavioral healthcare, Dr. Sharvit explores what makes generative AI different from earlier digital interventions and how it can support clinicians with documentation, assessments, continuous support, and care between sessions. She argues that human connection, clinical judgment, and understanding people in context must remain central even as more routine work is delegated to AI.

In this episode of Beyond the Couch: AI and Psychology, Dr. Ernest Wayde sits down with Dr. Hiba Arnaout, a postdoctoral researcher at the Ubiquitous Knowledge Processing (UKP) Lab at TU Darmstadt. Dr. Arnaout’s research focuses on natural language processing, knowledge graphs, digital mental health, scholarly AI, and responsible AI. She discusses how AI is already being used for mental health support, particularly when people face barriers to traditional therapy, and explores how AI could help improve access and support therapists with tasks such as therapy session summaries and clinical insights. The conversation also examines the risks of AI in mental health, including hallucinations, unreliable information, and challenges in evaluating AI-generated outputs. Dr. Arnaout discusses the growing use of AI as a therapist or companion and emphasizes the importance of systematic evaluation, human oversight, and collaboration between psychologists and computer scientists to develop safer, clinically grounded AI tools.

In this episode of Beyond the Couch: AI in Psychology, Dr. Ernest Wayde examines what happens when AI capabilities and adoption move faster than the systems designed to govern them. Drawing connections between recent AI cybersecurity testing, rapid AI adoption in healthcare, shadow AI, and evolving clinical workflows, he argues that responsible AI governance must extend beyond vendor approval, security reviews, and initial validation. Organizations also need to understand what AI systems can access, what actions they can take, how their capabilities change, and how unexpected behaviors will be identified after deployment. Dr. Wayde also explores a less visible dimension of AI risk: how repeated reliance on artificial intelligence may change human judgment, skills, and professional practice over time. For psychologists and other healthcare professionals, this raises important questions about clinical reasoning, interviewing, risk assessment, training, and the expertise needed to recognize when an AI-generated recommendation does not fit the situation. The goal is not to stop AI adoption, but to build ongoing governance that captures its benefits while continuously monitoring risks to patients, professionals, workflows, and the capabilities humans still need to preserve.