From the CMIO Seat: Principles That Matter for AI in Radiology
AI in radiology has no shortage of opinions. But when there’s consistent alignment, we should pay careful attention and listen, especially from leaders responsible for clinical care, innovation, and real-world deployment.
Across a series of conversations, three Chief Medical Information Officers at Rad AI–Elizabeth Bergey, MD, Andrew Del Gaizo, MD, and Rishi Seth, MD, CIIP, approached AI from different vantage points. Yet despite those differences, they consistently converged on three core principles.
Here’s what they all agree on.
AI Must Empower Radiologists, Not Add Friction
As hype around AI begins to settle and expectations around proof and ROI rise, the solutions that will endure are the ones that genuinely empower radiologists in their daily work.
Dr. Elizabeth Bergey explains where AI can make the biggest difference. She notes that AI’s real value lies in its ability to “amplify the value radiologists bring to patient care by freeing up time currently consumed by clerical duties.” When that burden is reduced, she explains, radiologists experience less strain and regain the capacity for “more meaningful consultative interactions with referring physicians.”
Dr. Rishi Seth echoes this perspective, emphasizing that traction comes not from novelty, but from removing friction where it actually exists. AI, he argues, should “surface priors, review documentation, structure findings, and draft clear impressions,” so radiologists can spend more time on image interpretation and clinical judgment. Too many tools, he cautions, solve interesting problems without truly saving time or reducing mental burden.
For Dr. Andrew Del Gaizo, the standard is even simpler. Any AI solution must pass a straightforward litmus test: “Does this tool give me time back to think and talk to clinicians and patients, or does it make me feel like a cog in the machine?” If it’s the latter, he says, he won’t use it himself, and he won’t authorize deploying it as a CMIO.
AI earns its place only when it removes burden. Built well, it frees radiologists from clerical tasks and repetitive actions, allowing them to spend more time on interpretation, clinical judgment, and patient care. Built poorly, it adds clicks, slows workflows, and introduces friction that ultimately undermines both adoption and trust.
How to Decide When to Build and When to Buy
As AI becomes more widely adopted, the bigger risk is no longer falling behind, but adopting technology that doesn’t align with how a practice actually works.
Dr. Andrew Del Gaizo emphasizes that buying makes sense when a problem is common and regulated, and “requires enterprise-grade uptime, integrations, support, and liability frameworks.” Building, by contrast, should be reserved for situations where workflows are truly differentiating and the organization can sustain long-term model operations, validation, and security.
Dr. Rishi Seth echoes this distinction. He advises practices to “buy when the solution is well-integrated and delivers quick ROI,” but to build when a capability is “unique or strategically differentiating.” In his view, the decision often comes down to data access, iteration speed, and who will own the workflow over time.
Dr. Elizabeth Bergey’s experience shows how this decision plays out in the real world. When her group was searching for a workflow orchestrator, she found that “every option we found was essentially a highly filtered conventional worklist.” The available buy options didn’t meet their needs. Instead, they required “a dynamic solution that could adapt to the craziness of everyday radiology,” accounting for shifting volumes, busy emergency departments, and physician availability. Rather than forcing a poor fit, her team conducted a make analysis and ultimately chose to build a custom solution designed to keep radiologists in their flow state.
Build versus buy isn’t about following the market. It’s about understanding your practice’s workflows well enough to choose technology that actually fits.
What True AI Success Looks Like in Radiology
Looking into the future, if AI becomes the hero of the healthcare story, then something has missed the mark. True success isn’t measured by how visible the technology is, but by how well it restores focus, purpose, and joy to radiologists.
For Dr. Rishi Seth, that future starts with AI disappearing into the workflow. He believes AI should “seamlessly exist within our PACS and reporting applications,” functioning as an intuitive extension of the radiologist rather than a separate system to manage. When AI works this way, radiologists don’t think about the technology at all, they simply “wonder how they ever worked without that ambient, intelligent assistant.”
Dr. Andrew Del Gaizo envisions a world where AI helps restore the joy of practicing medicine. “If we spend less time hunting for data and performing administrative tasks,” he explains, “there is tremendous potential to reduce the burnout symptoms many doctors experience.” That motivation is personal. “This is why I leaned into the CMIO and entrepreneurial path,” he adds, “to create tools that give radiologists their time and their voice back. When we do that, the altruism that drew many of us to medicine can shine through.”
For Dr. Elizabeth Bergey, success comes down to removing repetitive work from the radiologist’s day. She hopes AI will “premeasure lesions, compare measurements to prior studies, generate tables or graphs that show changes over time, and draft reports within each radiologist’s preferred template.” When those repetitive tasks are handled in the background, she concludes, “radiologists will finally be free to focus on what truly matters: clinical interpretation and patient care.”
In the future these CMIOs envision, AI works quietly in the background, taking on repetitive clerical work and clearing space for what matters most. The spotlight stays where it belongs: on radiologists and the care they provide.
Putting Radiologists Back at the Center
Taken together, these perspectives offer a grounded framework for how healthcare leaders should think about AI today. Not as a race to adopt the latest technology, but as a disciplined effort to support radiologists.
When AI empowers rather than distracts, fits rather than forces change, and fades into the background rather than demanding attention, it does what it was meant to do. It gives radiologists back time, clarity, and agency, and with it, the ability to focus on the work that drew many of them to medicine in the first place.
Continuing the Conversation
This perspective builds on our ongoing CMIO Q&A series. In our latest installment, Dr. Elizabeth Bergey shares how retention, fairness, and long-term thinking guide her approach to sustaining radiology teams. Read here.
