AI Is No Longer Optional in Healthcare — It’s Infrastructure
Healthcare is at a tipping point. Clinicians are overwhelmed, burnout is rising, administrative tasks are taking up too much time and shortages are rising — with the U.S. projected to face a shortage of ~ 187,000 physicians by 2037 and nursing shortages potentially exceeding 200,000 openings by 2030.
The good news: AI is the answer to combating many of these challenges.
In a recent episode of “Care Shift” with Sathvik Bilakanti, Rad AI Chief Innovation Officer Demetri Giannikopoulos shared how these challenges can be addressed with the support of AI, moving AI beyond a “nice-to-have tool” and evolving into core clinical infrastructure.
Here are highlights from the conversation.
The Workforce Crisis Is Driving AI Adoption
Healthcare isn’t preparing for a future shortage — it’s already in one.
“The workforce just can’t keep up,” said Gianniakopoulos. “AI is one of those few tools that actually allow us to help clinicians, help nurses, help patients manage their care, while still maintaining that really high quality of care”
Even with the benefits, a persistent concern in healthcare is whether AI will replace jobs. Rather than replacing clinicians, AI is emerging as a pressure-release valve.
“We’re in the midst of a crisis today and AI is the pressure release valve on that entire system,” he said.
So, what does this mean for radiologists?
Gianniakopoulos quoted Stanford Medicine Professor of Radiology Curtis Langlotz, MD PhD, saying radiologists that use AI will replace those that don’t.
The implication is clear: AI is becoming table stakes. Those who adopt it gain efficiency, clarity and time — those who don’t risk falling behind.
Radiology: The Frontline of Clinical AI
Among all specialties, radiology has emerged as the proving ground for AI in healthcare. “Radiology is easily the most mature clinical entry point for artificial intelligence,” said Gianniakopoulos.
Radiology has long been a leader in digital transformation, from early PACS systems to structured reporting. That foundation makes it uniquely suited for AI-driven workflows.
But, more importantly, radiology illustrates what’s possible across all specialties, including one of its most immediate benefits: time-savings.
Primary care visits often last just 15–20 minutes — time that must cover diagnosis, conversation and documentation. “Giving the physician literally time back to do exactly what they trained for, which is care for the patient, is where AI can fit into that,” said Gianniakopoulos.
In radiology, this manifests through automation of reporting and reduction of cognitive load. “We’re not talking about saving five seconds a day. We’re talking about hours over the course of a month,” he said.
That time compounds into faster diagnoses, quicker turnaround and better patient experiences.
Mission-Critical Tech Requires Trust
Healthcare organizations are also becoming more discerning about AI. Flashy demos and vague ROI claims are no longer cutting it.
Today’s buyers expect solutions that integrate into mission-critical workflows. “Being able to address systemic, mission-critical workflows is where there’s a lot of interest right now,” said Gianniakopolous.
This shift is pushing AI vendors to think differently — not just about features but about reliability, scalability and trust, with trust underpinning everything.
“Trust is the currency that determines whether or not the technology is going to become the standard of care,” he said
Building that trust requires more than accuracy metrics. It demands:
- Transparency in how models work
- Alignment with clinician workflows
- Personalization to individual users
Rad AI, for example, builds models tailored to each radiologist’s voice and style — helping clinicians feel ownership over the output.
“They trusted it because it was their voice,” said Gianniakopoulos. “What you really have to differentiate on is your understanding of the system, your understanding of the users,” he said.
The Future: 40–50% Efficiency Gains
Looking ahead, the impact of AI could be transformative, and Gianniakopoulos said the innovation expectations are high.
“Agentic is absolutely one of our focus points for the next couple years because there’s the ability to build infrastructure upon which you can just implement these incremental updates back up to making huge updates,” he said. “We don’t need to get 10-15% efficiency at the physician level. We need to get 40-50% efficiency … to ultimately achieve the outcomes [healthcare systems need].”
That level of efficiency doesn’t just improve workflows. It reshapes access to care, through faster reads, shorter wait times and earlier diagnoses.
Why This Moment Matters
The combination of workforce shortages, rising demand and administrative overload have strained the healthcare system. As Gianniakopoulos puts it, “AI is infrastructure … this is now central to care.”
The organizations that embrace this shift thoughtfully — prioritizing trust, workflow integration and real clinical value — will define the next era of healthcare.
