Eyeworld

WINTER 2026 | EYEWORLD | 85 P Relevant disclosures Barkey: Bausch Surgical Jacobs: None Rosenberg: None Seals: None Wortz: MDConsult AI Contact Barkey: patti.barkey@bausch.com Jacobs: Carrie.Jacobs@chuvision.com Rosenberg: ericr29@gmail.com Seals: bens@thomaseye.com Wortz: 2020md@gmail.com • How does it handle PHI, and is it HIPAA-com- pliant? • What are the exit terms if it doesn't perform? • Is the team ready, or does this require training? When is the right time to adopt technology? Mr. Seals said as a practice, you must decide where on the spectrum you want to be. He said there are "three different positions that you can take": • The leading edge: Early pain, competitive advantage, trailblazer status • The fast follower: Lower risk, proven tools, still ahead of laggards • The late adopter: Minimal disruption now, but missed margin opportunities With technology, you have the opportuni- ty to find the right place for your practice and where you want to be. The technology is changing so rapidly, Ms. Barkey said, adding that you need to look at what it's going to do for your practice and how much you're going to have to pick up after it. Dr. Wortz likened adopting some of these technologies to "waiting for the perfect wave to come." You can wait all day, but you might miss a lot, and you also see other people catching waves. At a certain point, you have to paddle and go, he said. It's about presenting an opportunity, then presenting risk and reward, he said, and part of that risk is time and investment. If it doesn't work, is there a downside? But if you see that "promised land" on the other side, that's the motivating factor to get in the game. You also have to understand what your product is going to impact. Whether it's a pa- tient-facing, doctor-facing, or payer-facing tool, there are different risks, Mr. Seals added. Specifically commenting on prompt engi- neering, Dr. Rosenberg suggested asking one question at a time and giving specific context (for example: the specialty, patient type, or goal). You want to be specific on every point that's going to be valuable to you as the provid- er, he said. "One of the ways I've gotten the best high-quality answers is not just with 'Here's what I want,' … It's going back and forth with multiple inputs and outputs to fine-tune the question," Dr. Wortz said. Once AI understands what you want after a number of questions, it will give you a high-quality output. Ms. Barkey also stressed the importance of careful question wording, adding that this has caused frustration for her. If you ask too many questions, over time, it will take you full circle, so you may need to tell it to stop or start over. "I 100% believe that if you aren't doing anything with AI right now, you need to start," she said, adding that it's also important to start introducing it to your staff. "We need to elimi- nate the fear of it replacing jobs so we can get people to embrace this." Concluding the session, some of the key takeaways were: • Strategy first: Technology serves your plan, not the other way around • Vet rigorously: Framework—assess fit, pilot with humans in the loop, exit fast if needed • Use the five pillars: Segment by workflow to find your lowest-risk, highest-reward entry point • Start simple: DIY tools, prompt engineering, and consumer AI are accessible today Aii glossary of terms and definitions Agent: An AI that can reason, plan, and take multi-step actions autonomously AGI (Artificial General Intelligence): Human-level reasoning (not here yet) Bot: A rule-based automation that executes a specific task Prompt: The instruction you give an AI— quality in = quality out LLM (Large Language Model): The engine behind ChatGPT, Gemini, Copilot RAG (Retrieval-Augmented Generation): AI that pulls from your data RPA (Robotic Process Automation): Technology that uses "bots" to mimic human actions Ms. Barkey, Ms. Jacobs, Dr. Rosenberg, Mr. Seals, and Dr. Wortz spoke at the Aii session on the adoption of AI and other technologies. Source: ASCRS

• Cover