Artificial intelligence in addiction medicine with Patrick KellyArtificial intelligence in addiction medicine with Patrick Kelly
Addiction Audio
Dr Elle Wadsworth talks with researcher Patrick Kelly about how clinicians and people who use drugs view artificial intelligence in addiction medicine. They discuss practical benefits, ethical risks, stigma, and why cautious optimism depends on transparency and genuine community involvement.
20:48•18 Sept 2026
Is Artificial Intelligence a Friend or Foe in Addiction Medicine?
Episode Overview
- AI is already used to diagnose substance use disorders, predict overdose risk, and support treatment decisions through clinical data.
- Clinicians value AI’s practical benefits but want transparent development and strong safeguards against reinforcing drug-related stigma.
- A fast-changing and regionally varied drug supply raises doubts about how long AI tools remain accurate and clinically useful.
- Many people who use drugs have limited knowledge of AI, often view it as untrustworthy, and prioritise human connection in care.
- People who use drugs may share medical data to help others, but they want control over identifiable information and protection from criminalisation.
“There is real momentum here when it comes to ai and addiction medicine, but it cannot be at the expense of the people that these tools are ultimately aiming to help.”
Curious about how others navigate their sobriety journey when technology gets involved? This Addiction Audio instalment takes a sharp look at artificial intelligence (AI) in addiction medicine, with a clear focus on what both clinicians and people who use drugs actually think about it. Host Dr Elle Wadsworth chats with Patrick Kelly, a doctoral candidate at Brown University School of Public Health, about his qualitative study pairing these two groups’ perspectives.
Patrick starts by framing AI as "almost like a giant toolkit" used for predictions and inferences, already helping to detect substance use disorders, predict overdose risk, and flag emerging harms through electronic health records and geospatial data. Clinicians in the study see practical benefits: AI tools can ease admin work, support diagnosis, and help them keep up with a rapidly changing drug supply. But their enthusiasm comes with conditions.
They want "transparent development" and strong reassurances that AI won’t bake more stigma into care for people who use drugs, especially given drug use is often criminalised. On the other side, many people who use drugs were surprised to learn how much AI is already used in healthcare and tended to view it as "potentially just really untrustworthy". They stressed that no algorithm can replace human compassion in the context of overdose risk and community grief.
Yet, interestingly, participants were often willing to share their medical data "altruistically" to improve care for others, provided they keep control over identifiable information and aren’t exposed to greater risk of criminalisation. Patrick finishes in a "cautiously optimistic" space, arguing that ethical data stewardship, community-rooted tool design, and better communication about what AI does and doesn’t do are essential.
As he puts it, "there is real momentum here when it comes to ai and addiction medicine, but it cannot be at the expense of the people that these tools are ultimately aiming to help." If AI is going to play a bigger role in addiction care, are we ready to make sure it truly works for the people most affected?
Do you want to link to this podcast?
Get the buttons here!
More From This Show
The latest episodes from the same podcast.