Pharmacists, Don’t Fear AI: ASHP Framework Shows How to Embrace AI

Steven Smoke, PharmD, MS, a clinical informatics pharmacist at RWJ Barnabas Health, has a simple message for pharmacists: Artificial intelligence (AI) is here to stay, and you can use it to do more than you think.

Speaking to a large, packed room at the American Society of Health-System Pharmacists Midyear 2025 Clinical Meeting and Exposition, Smoke laid out AI's potential and shortcomings in pharmacy practice and shared his 2-dimensional framework for evaluating whether AI use is appropriate for a given situation.

AI Hype vs AI Slop

When Smoke asked the room to share their experience with and thoughts on AI, it revealed a healthy mix of people who use it regularly, those who have never used it, and people who are excited, "cautiously optimistic," or fearful of AI's potential. He also asked the room to share how often they used AI; most answered either daily or several times a week. This provided the baseline for the rest of Smoke's presentation and let him know what concerns were relevant for him to address.

Smoke said he categorizes AI outputs as either "AI hype" (output that performs well and raises the quality of work) or "AI slop" ("low-quality, low-effort uses of AI," as Smoke said). He provided real-world pharmacy examples of both: "[There] are studies where AI is used to read basic imaging studies, like chest X-rays and CT scans, and drawing incredible conclusions like diagnosing diabetes, heart failure, and cardiovascular risk," Smoke said. He also cited AI's use in drug development, where tools "are transforming our understanding of protein structure and binding." This, he said, is AI hype, or AI performing at its best.

On the AI slop side, Smoke described "an AI algorithm used by insurance to predict patients' health care costs and needs, and it was found to be racially biased." He also cited a well-documented case of a patient diagnosed with bromism after an AI chatbot recommended using sodium bromide as a replacement for table salt.

"It's not hard to see how...you can draw very different conclusions about what AI is going to mean for medicine [and] pharmacy," Smoke said.

How Should Pharmacists Evaluate the Appropriateness of AI?

To help pharmacists determine, on both individual and system-wide levels, whether AI should be used in a particular situation, Smoke said to ask 2 questions: Am I asking AI to provide a predictive output or to generate an original output? Is this function near or far from the patient?

The difference between generative and predictive AI is crucial, and the 2 are quite distinct from each other. "Predictive AI is AI that is designed to perform a single, specific task," Smoke said, and the tool should have a large body of data to draw upon when answering the question and be able to predict an unambiguous outcome. He used the example of asking AI to determine inventory needs in a pharmacy practice, which would have large amounts of past purchasing, census, and use rates data to draw upon and be able to give an unambiguous answer.

"On the other hand, generative AI is AI that needs to generate something," Smoke said. "The training data for tools like this...[is] often very general."

For this, he used the example of drafting a note for an electronic medical record. "If you're spending a lot of time drafting notes in charts, then this would certainly be [useful]," he said, but it would still need to be reviewed by a pharmacist or other professional. The distinction between these and other types of AI is important, Smoke said, because "AI is such a broad umbrella term. If you're talking to someone about AI, and you're thinking about something like a sepsis [prediction] tool and they're thinking about [a] chatbot for a patient, you're having 2 [different] conversations."

The second question to ask is whether the AI function is "near" to or "far" from the patient. Tools near the patient are anything that impacts patient care directly or requires specific patient information. "Tools that are further from them [are] quality assessments, operations, billing, [and] administrative aspects," he said. "This assessment is important because it helps us understand the layers of safety between AI and patient care, and that will really impact our risk and threshold for safety as we assess these tools."

By using the framework he outlined, Smoke said pharmacists and health systems can facilitate meaningful conversations about AI tools.

Pharmacy's Role in the AI Era

Smoke said he doesn't want to predict any timeline for AI's use in the industry to become widespread, let alone when AI would do the thing some in his audience feared and take pharmacy jobs. But he praised pharmacists' long history of adaptability and of evolving as new roles and new skills became essential to the practice. AI, he said, is likely to become one of these skills. "We need to know how to work with AI," he said. "We don't need to know how to [build] AI."

Pharmacists still have an essential role to play in the AI landscape, especially in oversight, quality assurance, and ensuring the ethical integration and use of AI within AI workflows. They are also essential to "interdisciplinary collaborations and working with data scientists and AI developers to make sure [this technology] can be used practically," he said.

Smoke gave a practical list of skills for pharmacists to develop for an AI-enhanced practice, beginning with his 2-dimensional framework and ending with continued familiarity with best practices for generative AI tools, especially as they continue to develop. "I would [also] encourage folks to do a deeper dive into different types of assessment and metrics" as a way to grow their ability to evaluate AI tools.

Most of all, he said, using AI is "the best way to get a healthy sense of what these things are good for and what they're not good for, especially as they continue to [evolve]. ... The better we can harness the power of AI while using [it] safely and effectively, the better we can use it to improve patient care and pharmacy operations."

REFERENCES

Smoke, Steven, PharmD, MS. Navigating the artificial intelligence (AI) revolution: applications, evaluation, and the pharmacist's role. Presented at: American Society of Health-System Pharmacists Midyear 2025 Clinical Meeting and Exposition; December 7-10, 2025; Las Vegas, Nevada.