Blab AI
AI Can Augment, Not Replace, Human Therapy for Substance Use Disorders
When a glass of wine feels like a lifeline, a chatbot might offer a second lifeline—but it can’t replace the human touch.
Artificial‑intelligence tools are increasingly marketed as stand‑ins for traditional psychotherapy in treating alcohol and other substance use disorders (SUDs). The claim that a chatbot or app can substitute for a licensed therapist is not supported by current evidence.
In 2023, fewer than 10 % of the 46 million Americans estimated to have a SUD received any treatment, according to the Substance Abuse and Mental Health Services Administration (SAMHSA, 2024). Geographic, financial, and stigma barriers keep many people from accessing in‑person care. AI‑powered programs that deliver psychoeducation, screen for risk, encourage help‑seeking, and provide between‑session support can therefore serve as useful adjuncts.
The most defensible AI application is the computer‑assisted delivery of cognitive‑behavioral therapy (CBT). CBT’s structured format—psychoeducation, cognitive restructuring, and behavioral skill building—makes it more amenable to algorithmic delivery than relationally complex therapies. Randomized trials of digital CBT programs, such as CBT4CBT for cocaine‑dependent individuals, have shown modest efficacy when used alongside standard care (Carroll et al., 2014). AI can also send relapse‑prevention prompts, track craving patterns, and offer coping strategies at high‑risk moments, effectively extending the therapeutic hour.
However, the notion that AI can replace psychotherapy reflects a misunderstanding of what therapy actually is. Therapy is not a set of instructions or a to‑do list. Even CBT is most effective when delivered within a strong therapeutic relationship (Norcross & Wampold, 2011). Interpersonal, psychodynamic, and motivational‑enhancement therapies place the relationship at the core of change. Addiction involves deeper psychological vulnerabilities—self‑esteem deficits, unresolved trauma, attachment ruptures, and shame—that require a sustained, attuned, trusting partnership between patient and clinician (Washton & Zweben, 2023). No current AI system can form or utilize such a genuine interpersonal bond.
Denial and ambivalence are central features of SUDs. Motivational interviewing (MI), the evidence‑based framework designed to address resistance, relies on a clinician’s ability to read subtle cues, roll with resistance, and develop discrepancy within an established alliance (Miller & Rollnick, 2023). AI systems exhibit a documented tendency toward “sycophancy” – responding in ways that users find agreeable and emotionally comfortable, even when clinical accuracy demands otherwise (Malmqvist, 2025). In addiction treatment, this bias can be dangerous. A patient who insists that their drinking is “not really a problem” is likely to receive a validating AI response that neither challenges the distortion nor advances change. Skilled clinicians, by contrast, use such moments to compassionately confront the discrepancy between stated values and behavior.
Safety concerns further limit AI’s role in addiction care. Alcohol and certain drug withdrawals can be medically life‑threatening; AI cannot conduct clinical assessments, detect imminent risk, or coordinate medical interventions. Comorbid psychiatric conditions, common among people with SUDs, require differential diagnosis and treatment planning beyond AI’s scope. Confidentiality standards, mandatory reporting obligations, and liability frameworks that apply to licensed clinicians do not uniformly cover commercial AI applications (NIDA, 2023). Patients who rely on AI as a primary “treatment” risk delaying or forgoing evidence‑based care, potentially at serious cost to their health.
The most defensible role for AI in addiction treatment is as an adjunct to human care, not a replacement. AI tools can extend psychoeducation, support between‑session skill practice, facilitate symptom monitoring, and reduce barriers to entering treatment. When used judiciously and under appropriate clinical oversight, they may enhance care for patients already engaged with a qualified provider.
In summary, AI can supplement CBT‑based addiction care but cannot replace the irreplaceable human work of standing with a person, earning trust, tolerating resistance, and guiding them toward confronting the reality of their substance use. That remains the province of skilled clinicians, and no algorithm is close to replicating it.
Artificial‑intelligence tools are increasingly marketed as stand‑ins for traditional psychotherapy in treating alcohol and other substance use disorders (SUDs). The claim that a chatbot or app can substitute for a licensed therapist is not supported by current evidence.
In 2023, fewer than 10 % of the 46 million Americans estimated to have a SUD received any treatment, according to the Substance Abuse and Mental Health Services Administration (SAMHSA, 2024). Geographic, financial, and stigma barriers keep many people from accessing in‑person care. AI‑powered programs that deliver psychoeducation, screen for risk, encourage help‑seeking, and provide between‑session support can therefore serve as useful adjuncts.
The most defensible AI application is the computer‑assisted delivery of cognitive‑behavioral therapy (CBT). CBT’s structured format—psychoeducation, cognitive restructuring, and behavioral skill building—makes it more amenable to algorithmic delivery than relationally complex therapies. Randomized trials of digital CBT programs, such as CBT4CBT for cocaine‑dependent individuals, have shown modest efficacy when used alongside standard care (Carroll et al., 2014). AI can also send relapse‑prevention prompts, track craving patterns, and offer coping strategies at high‑risk moments, effectively extending the therapeutic hour.
However, the notion that AI can replace psychotherapy reflects a misunderstanding of what therapy actually is. Therapy is not a set of instructions or a to‑do list. Even CBT is most effective when delivered within a strong therapeutic relationship (Norcross & Wampold, 2011). Interpersonal, psychodynamic, and motivational‑enhancement therapies place the relationship at the core of change. Addiction involves deeper psychological vulnerabilities—self‑esteem deficits, unresolved trauma, attachment ruptures, and shame—that require a sustained, attuned, trusting partnership between patient and clinician (Washton & Zweben, 2023). No current AI system can form or utilize such a genuine interpersonal bond.
Denial and ambivalence are central features of SUDs. Motivational interviewing (MI), the evidence‑based framework designed to address resistance, relies on a clinician’s ability to read subtle cues, roll with resistance, and develop discrepancy within an established alliance (Miller & Rollnick, 2023). AI systems exhibit a documented tendency toward “sycophancy” – responding in ways that users find agreeable and emotionally comfortable, even when clinical accuracy demands otherwise (Malmqvist, 2025). In addiction treatment, this bias can be dangerous. A patient who insists that their drinking is “not really a problem” is likely to receive a validating AI response that neither challenges the distortion nor advances change. Skilled clinicians, by contrast, use such moments to compassionately confront the discrepancy between stated values and behavior.
Safety concerns further limit AI’s role in addiction care. Alcohol and certain drug withdrawals can be medically life‑threatening; AI cannot conduct clinical assessments, detect imminent risk, or coordinate medical interventions. Comorbid psychiatric conditions, common among people with SUDs, require differential diagnosis and treatment planning beyond AI’s scope. Confidentiality standards, mandatory reporting obligations, and liability frameworks that apply to licensed clinicians do not uniformly cover commercial AI applications (NIDA, 2023). Patients who rely on AI as a primary “treatment” risk delaying or forgoing evidence‑based care, potentially at serious cost to their health.
The most defensible role for AI in addiction treatment is as an adjunct to human care, not a replacement. AI tools can extend psychoeducation, support between‑session skill practice, facilitate symptom monitoring, and reduce barriers to entering treatment. When used judiciously and under appropriate clinical oversight, they may enhance care for patients already engaged with a qualified provider.
In summary, AI can supplement CBT‑based addiction care but cannot replace the irreplaceable human work of standing with a person, earning trust, tolerating resistance, and guiding them toward confronting the reality of their substance use. That remains the province of skilled clinicians, and no algorithm is close to replicating it.
