Over the past 100 years, our mental model of a typical psychotherapy session has not changed much: You probably imagine a distraught but comfortable client lying supine on a couch while a curious therapist takes notes and ponders about how to respond. It is our prediction that with the meteoric rise of Large-Language-Model (LLM)-based artificial intelligence, this image will soon only be an amusing meme of a less precise and effective history when helping other human beings was more of an artform than a science.
To say that psychotherapy lacks any scientific rigor is certainly false. Since its inception, clinical scientists have published thousands of studies that examine the efficacy of various psychotherapies and their respective techniques that are purported to produce meaningful reductions in mental distress. Despite this wealth of scientific knowledge, psychotherapy has also been described as an art because the incredible variability of human behavior requires therapists to rely on frameworks as opposed to precise rules to manage a myriad of unpredictable situations. Unfortunately, however, the overreliance on the ‘artform’ of psychotherapy has led to a proliferation of ineffective treatments that don’t rely on scientific evidence, and consequently psychotherapy lags scientifically behind other therapeutic areas, like immunology and oncology, where data-driven decisions rule. Further, many mental health clinicians rely on intuition or lived experience instead of the flexible application of known treatment modalities.
So what are good therapists doing that results in positive change for those struggling with mental illness? Previously, the answer to this question was largely a mystery. While there was an abundance of evidence suggesting that psychotherapy is better than no treatment at all, and that therapy often performs equivalently to the use of medication, the specific words and their optimal configuration to produce change was largely unknown. Luckily, within the past 10 years our knowledge about the science of language intervention has improved dramatically because of the rapid growth of text-based psychotherapy, as well as telehealth — particularly during Covid-19. During this time, several large companies began to rely on text-based care as a novel therapeutic modality to provide care to more people, and also made therapy available via telehealth. Though initially questionable as a suitable equivalent to in-person face-to-face interaction, both text-based care and telehelath were soon discovered to produce equivalent treatment outcomes in most ways to face-to-face care. This shift also led to something amazing: a wealth of data of communications between patients and providers -a veritable treasure trove of data for the scientific community to uncover the mysteries of how words can be judiciously selected to produce just the right therapeutic outcome for a given patient at a given time. For example, we now know that as people get better they start to use language with a future verb tense instead of utilizing the present or past; knowledge which can now be harnessed to encourage a different orientation that leads to more rapid recovery from depression.
With these advances we are more acutely aware than ever that words are important. The right words are important. The right words are important at the right time.
A subtle “You can do this!” from a parent who helps their child to acquire a new skill, or a perfectly timed “I love you” are just two examples of how words can create transformative moments in our lives. This fact is truly incredible — that words can be both destructive (“I hate you” or “you’re fired”) as well as healing when used correctly and with precision. In the context of psychotherapy, words are very potent tools that are both harnessed to unlock potential and alleviate mental illness. Their application is incredibly nuanced, with personalization required for each patient (e.g., the same message must be customized to accommodate a different ethnicity, age, or lived experience), and a sophisticated knowledge of a patient’s history is required to understand when a patient is maximally receptive to hear the right words. For therapists using words to both assess the problem and to produce an outcome is a unique challenge. Psychotherapy is in fact unique in that it is the only field of medicine for which spoken language is both a primary diagnostic and a primary therapeutic tool.
While psychotherapy is arguably in its true renaissance period because of the aforementioned advances, we have now begun a new wave of acceleration in the science of language intervention: one enabled by LLM-based artificial intelligence (AI). Artificial Intelligence and machine learning have already accelerated this process of knowledge and discovery of language intervention to previously unimaginable levels of precision and personalization. We believe this will be even more impactful than advances we have experienced so far. Given that LLMs are effectively controlled via English (as their programming language), this provides a major opportunity for us to further advance our understanding of how to help. By safety and cautiously integrating LLMs in a clinician setting, we have the potential for the ultimate levels of personalization: the perfect words for the right person at the right time.
In the future, it is entirely possible that traditional psychotherapy as we know it may be changed as a new treatment modality that is routinely coupled with AI-powered technologies. It is also likely, however, that therapists will still be critical to the optimal recovery of those with mental health concerns. Though AI will likely be able to harness the power of selecting the correct word to use with the right person with the correct timing, human beings remain optimally suited to collaboratively craft and execute a “meta” plan with patients that takes into account the holistic experience and ultimate desires of the individual -often in ways that are not consciously available to the patient him/her/themselves. Consequently, a more immediate future state is probable where clinicians are augmented by AI in a way that produces ‘super clinicians’ like we’ve never encountered before. For example, clinicians will have new insights about diagnosis, increased awareness of risk factors, more real-time understanding of clients, and more effective and timely interventions that can be e-prescribed. It is also likely that AI will be utilized to help clients between sessions to make progress and to afford 24/7 access to high quality, meaningful interventions that are directed by the mental health clinician.
As we are all acutely aware, no one can accurately predict the future. What we think is near certain, however, is that psychotherapy has and will be changed forever with LLMs, whether we like it or not. Our hope and belief is that it will be for the better, and it is our duty to build and contribute toward that future, as opposed to avoiding it.
Photo: Vertigo3d, Getty Images
Bill Hudenko, Ph.D. has significant experience in the fields of both mental health and technology. Dr. Hudenko is a licensed psychologist, a researcher, and a professor who holds a joint appointment as a faculty member at Dartmouth’s Department of Psychological and Brain Sciences and Dartmouth’s Geisel School of Medicine. His research focuses on the use of technology to improve mental health delivery and patient outcomes. He has worked with hundreds of clients and has taught thousands of students during his tenure at Dartmouth, Cornell University, and Ithaca College. Dr. Hudenko is also an experienced entrepreneur and is the former CEO of Trusst Health Inc., Voi Inc. and Incente, LLC -all mental health tech startups designed to transform the delivery of mental healthcare through technology. Dr. Hudenko is currently the Chief Clinical Officer at Jimini Health, a company utilizing AI to augment human therapists.
Luis Voloch is the Co-founder and CEO of Jimini Health. Prior to Jimini Health, Luis co-founded and served as CTO of Immunai, an AI-driven drug discovery company that achieved a billion-dollar valuation with over 140 employees. A MIT alumnus with degrees in mathematics and computer science, Luis brings deep expertise in machine learning and biotech innovation. Luis won the best Thesis Award among all-PhD track students in MIT for EECS. His career spans leadership roles at Palantir and ITC, where he spearheaded data science and ML initiatives. Currently, he also lectures at Stanford Graduate School of Business, teaching entrepreneurship and management in AI-heavy companies. Luis is passionate about leveraging artificial intelligence to solve complex challenges in healthcare, from accelerating drug discovery to transforming mental health care delivery.
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.
