DENVER — AI seems like a natural fit for assisting in the visually driven specialty of dermatology, and it’s already transforming administrative and operational tasks. It holds great promise for streamlining skin disease detection, diagnosis, and prognosis, experts say. On the other hand, the technology is evolving faster than education and awareness of its safe and ethical use can keep up, which has raised concerns about its impact on training, the specialty, and patient care.

Using AI to augment dermoscopy, provide clinical decision support, and advance total body photography are among the pluses, said Yevgeniy “Eugene” R. Semenov, MD, PhD, assistant professor of dermatology at Harvard Medical School in Boston.

“This is a very hot topic now, but there are two sides, of course, to AI. What I’ll present today are the positives of it in dermatology,” Semenov said during a debate on AI here at the American Academy of Dermatology (AAD) 2026 Annual Meeting. “The ultimate goal is to be able to leverage some of these technologies that are becoming increasingly efficient and increasingly comprehensive to augment the way that we practice clinical care.”

However, there are also numerous unknowns, including unintentional consequences, and a fear that overreliance on AI could lead to skill depletion or “never skilling” among medical students and trainees, said Shannon Wongvibulsin, MD, PhD, dermatology resident in the University of California, Los Angeles Specialty Training and Advanced Research Program.

Wongvibulsin shared some cautionary examples, presenting an actual prompt a medical student used on OpenEvidence AI: “Rewrite this so it is professional and includes extra information showing that I looked stuff up.” This points to a great need for education on responsible, ethical, and safe use of these tools, she said.

In another case, part of an AI response was published in a journal, which included, “Here is a possible introduction for your topic, and I’m very sorry, but I don’t have access to real-time information or patient-specific data, as I am an AI language model.” Wongvibulsin added, “This got by the authors, reviewers, and the journal.”

While Semenov expects that in a year or 2, most dermatologists will be using AI in their daily practice, less than half currently believe AI will improve clinical care. “People are a little skeptical. I do believe that this number will hopefully go up with time, but we’ll have to prove that AI actually leads to better outcomes,” he said.

There are two commercially available, AI-enabled technologies approved by the FDA. One is the DermaSensor device to aid skin cancer detection in primary care settings. A second product, the Nevisense device, is designed to assist in early melanoma detection.

Picking the ‘Lowest Hanging Fruit’

The biggest inroads for AI in dermatology so far are assisting in image analysis, clinical decision support, and operations management. These are the “lowest hanging fruit,” Semenov said. AI patient scheduling is a prime example, where AI “essentially cuts out the human component, which can be a source of delay,” Semenov said. AI-driven ambient scribe technology also holds potential to reduce administrative burdens and lower physician burnout, authors of a 2025 JAMA Network Open paper pointed out.

“Ultimately, AI can improve clinical productivity and reduce physician burnout all through optimizing clinical workflows, improving patient scheduling, and producing documentation,” Semenov said.

AI can also tailor patient handouts, Semenov said. Examples include adding the following: What do you do after a surgery? What does your diagnosis mean? What is the most common treatment? AI can “maybe even send a letter to an insurance company that would then use their own AI to rebut it and deny a drug,” he said.

“But you know, this is the world we have now,” he said.

Will AI Erode Skills?

“Although the hope is that AI can help augment our abilities, it’s also essential to understand the risk of de-skilling and never skilling,” Wongvibulsin said. In fact, skill loss is clearly recognized as a risk of AI, especially among early career physicians, she added. In a 2026 American Medical Association Physician Survey on Augmented Intelligence, 88% of respondents indicated concern about skill loss.

De-skilling occurs when someone whose established skills degrade over time with AI use. For trainees who have not yet acquired these skills, there is a risk of “never skilling,” she pointed out.

They may never achieve proficiency in “their clinical skills, and instead, simply depend on these AI tools,” Wongvibulsin said. “But without this clinical skill set, how do we know when these tools are not performing properly?”

Another caveat: Supporting evidence and dermatologist involvement in the development of these tools, especially ones that are available in app stores, may be lacking, Wongvibulsin added. In addition, generative AI tools are trained to be helpful and tend to obsessively agree with users, often at the expense of accuracy, she said.

At the meeting, the debate over the pros and cons of AI in dermatology engaged several meeting attendees to ask questions, including from the A/V person in the room. He asked how involved dermatologists get in the development of these technologies vs software engineers.

“I think it’s both,” Semenov said. The engineers are developing the technology, “and then we’re adapting it to medicine.” He added that costs are a factor, with AI-driven models costing up to $20 million or more. “You can’t do this in healthcare alone. Individual healthcare centers don’t have the bandwidth to do this.”

But, he added, “Ultimately, we are responding, and we do control the data, which is a good thing.”

Unknowns Remain

Although these tools are rapidly proliferating, education in the space is not occurring at the same pace, Wongvibulsin said. “Further research is still necessary to understand the impact of these tools and the performance in real-world settings.”

“There are still a lot of unknowns in this space, [including] a lot of unknowns about potential unintended consequences of these tools,” Wongvibulsin said. AI technology should be transparent, for example, on the range of skin tones on which the system was trained.

“And concerningly, the most dangerous advice can come from the most well intentioned of chatbots,” she added.

Session director Kenneth Tsai, MD, PhD, agreed that some caution is warranted. When asked to comment, he told Medscape Medical News, “Frankly, the deployment of AI in dermatology seems rather uneven and fractured. It’s important for dermatologists to be very intentional about using it in practice, be aware of the pitfalls, and always incorporate some degree of quality control.” Tsai is a dermatologist and dermatopathologist at Moffitt Cancer Center in Tampa, Florida, specializing in high-risk skin cancers.

More information on AI and dermatology is available from the American Academy of Dermatology and the British Association of Dermatologists joint consensus statement on AI.

Semenov and Wongvibulsin reported having no relevant disclosures for their presentations.

Semenov disclosed being a consultant for Alterome Therapeutics, Arcutis Biotherapeutics, Bristol Myers Squibb, Pfizer, and Pyxis Oncology and being a member of advisory boards for Galderma, Regeneron, and Sanofi. Wongvibulsin disclosed being a consultant for VisualDx and a member of the advisory board for Sanofi and Regeneron. Tsai disclosed being a consultant for Altos Labs, DX Biosciences, and NFlection Therapeutics; being an independent contractor for Ankyra Therapeutics, APIS Assay Technologies, UpToDate, and Verrica Pharmaceuticals; being an investigator for Incyte; being a member of the advisory board for Regeneron; and being a stockholder in 3CC-Third Culture Capital.

Damian McNamara is a freelance contributor to Medscape Medical News. He worked full-time for Medscape and WebMD from 2018 to 2024. Damian has a BA in chemistry and an MA in science, health, and environmental reporting or journalism.