Table of Contents
At a time when the mental health sector is overwhelmed and under-resourced, researchers at Dartmouth College have spent six years developing Therabot, an artificial intelligence-powered psychotherapy application. Unlike many of the mental health apps saturating today’s market, Therabot is designed with a firm foundation in clinical science—and it’s already showing promising results.
“There simply aren’t enough therapists. Even if we multiplied the current workforce by ten, we’d still fall short,” said Nick Jacobson, assistant professor of data science and psychiatry at Dartmouth. “We need something fundamentally different to meet this large and growing need.”
Backed by Research, Not Just Hype
The Dartmouth team has published a clinical study demonstrating Therabot’s ability to help users manage anxiety, depression, and eating disorders. The results, which are currently being prepared for a larger comparative trial against conventional therapy, offer rare credibility in a field where many apps make bold claims without scientific backing.
Jacobson’s colleague, psychiatrist Michael Heinz, emphasized the team’s deliberate pace. “We could have rushed this to market like others have,” he said, “but safety, trust, and clinical effectiveness have always come first.”
The team is even considering launching a nonprofit to ensure Therabot reaches those who cannot afford traditional mental health services, reaffirming their mission of care over commerce.
Therabot vs. the Mental Health App Market
Therabot’s cautious development contrasts sharply with the often unregulated mental health app ecosystem, where many platforms aim more to retain user attention than to heal.
Vaile Wright, senior director of health care innovation at the American Psychological Association (APA), sees potential: “We envision a future where AI chatbots, rooted in science and co-developed by experts, support mental health care responsibly.” Still, Wright voiced concern over apps that may cause harm, especially to younger users.
Many such platforms are optimized to say what users want to hear, rather than what they need—especially problematic when their audiences are vulnerable teens and young adults.
Manual Training Sets Therabot Apart
Rather than scraping therapy transcripts or relying solely on training data from online sources, the Therabot team manually crafted simulated therapist-patient dialogues to train the AI. This uncommon step was taken to maintain control over quality, context, and emotional tone.
Their method stands in contrast to apps developed using mined content, which may reinforce biases or miss key therapeutic nuances.
No Clear FDA Oversight—Yet
Despite the U.S. Food and Drug Administration (FDA) being nominally responsible for digital health products, including AI tools, there’s little direct regulation of mental health apps.
According to an FDA spokesperson, such apps aren’t certified as medical devices but may be authorized for marketing after a formal pre-market review. The FDA did acknowledge the promise of digital therapies in expanding access to behavioral health care.
The Always-On Advantage—and Its Limits
Advocates of AI mental health support highlight a key strength: availability. Human therapists can’t be on call 24/7, but apps like Therabot or Panda (by startup Earkick) offer constant accessibility.
Grip Strength May Predict Lifespan and Overall Health, Experts Say
“Calling your therapist at two in the morning is just not possible,” said Herbert Bay, CEO of Earkick. “But a therapy chatbot is always there.” Bay also stressed that Panda is being clinically tested for crisis detection, including signs of suicidal ideation, and includes built-in alert systems.
Bay referenced a disturbing case involving Character.AI, where a mother blamed the chatbot for influencing her 14-year-old son’s suicide. “That can’t happen with us,” he said, underscoring the need for robust safety measures.
Public Voices: Some Find Help in Existing AI
Despite not being designed for therapeutic use, even general-purpose AI tools like ChatGPT are being used by people for mental health support.
A user named Darren, who suffers from traumatic stress disorder, said that interacting with ChatGPT had helped him manage anxiety. “I feel like it’s working for me,” he said. “I would recommend it to people in distress.”
This grassroots use of AI reflects both the desperation for accessible care and the early signs that machine-driven conversations can hold therapeutic value—provided they’re handled responsibly.
A Tipping Point for Mental Health Technology
Experts agree: while AI is unlikely to replace trained therapists anytime soon, it can play a critical supporting role in daily mental health maintenance and early intervention. The key lies in building trustworthy, evidence-backed, and ethically developed tools like Therabot.
“We’re not claiming this is a silver bullet,” said Jacobson. “But it’s a strong, science-based start—and it might just help fill a gap that the current system cannot.”
