AI’s role in healthcare is no longer a question of if but how.
In a Sept. 10 keynote at the Oracle Health and Life Sciences Summit in Orlando, Oracle Industries President Mike Sicilia and OpenAI Vice President of Health Nate Gross, MD, outlined a vision for technology that empowers patients while candidly addressing its limitations.
A partnership in the spotlight
The keynote session unfolded against the backdrop of a historic week for Oracle: Reuters reported the company’s valuation surged to nearly $1 trillion after announcing a 359% increase in future contract revenue — gains fueled in part by a $300 billion, multi-year computing partnership with OpenAI.
The tie-up is already extending into healthcare, with Oracle announcing plans to embed OpenAI technology into its patient portal.
Mr. Sicilia demonstrated the new AI-powered portal during the session, showing how patients can view lab results in plain language, receive personalized question prompts ahead of doctor visits and even explore clinical trial opportunities. Patients will also be able to ask questions about their care and receive recommendations for next steps, as well as the ability to connect with their care team if needed. He stressed that safeguards are built in and the AI shows its work.
“Every single piece of information that’s generated by the model is completely cited. It’s completely auditable, so that providers — and patients as well, should they choose to do so — see exactly how that decision was made,” he said.
Empowering patients through AI
For Dr. Gross — who joined OpenAI in June after co-founding both Doximity, where he spent 15 years, and Rock Health, where he remains a board member — healthcare represents one of the clearest ways to fulfill the company’s mission of ensuring AI benefits everyone.
“Health is going to be one of the purest ways to deliver on that mission,” he said, citing that nearly 9 in 10 U.S. adults struggle with health literacy, which undermines their ability to navigate the healthcare system. “It’s going to help patients take a more active role in their care.”
He pointed to opportunities for AI to help patients decode diagnoses, prepare for increasingly compressed care visits and adopt healthy routines that could reduce unnecessary care and costs. Still, challenges remain. From hallucinations to integration hurdles, Dr. Gross cautioned that success requires close collaboration with providers.
“To get to the right place in healthcare, we always need to look at AI as a means to an end, not an end in itself,” Dr. Gross said.
OpenAI has been active in making sure their models are useful and trustworthy for healthcare. The trust from clinicians, researchers and patients is paramount for adoption, and in healthcare there is little margin for error. It’s different from other industries, where failures and interaction are common; failures have severe consequences in healthcare.
“We’re not trying to shove AI into healthcare. It’s very much been a pull,” said Dr. Gross. “Health questions are a top use case in ChatGPT, and so we believe that creates a responsibility on our side to ensure that we have the safe, accurate and genuinely helpful responses.”
To engender trust, the company released HealthBench, a physician-led rubric evaluation that helps solve the issues with AI as they arise. The physicians can check whether the AI models made the right choice in real-world scenarios that mirror how individuals and clinicians actually interact with the platform. OpenAI took the learnings to improve training with the evaluations.
“We’ve seen dramatic progress in health-related needs in GPT. That gives me confidence that we’re moving in the right direction, but it also reinforces the thoughtful, co-dependent-cobuilding is the way to achieve our goals,” said Dr. Gross.
Beyond the ‘longitudinal record’
Both leaders also took aim at the shortcomings of current health data.
“The longitudinal record is not actually right — it’s often a confederacy of data,” Dr. Gross said.
Mr. Sicilia echoed the point, adding that true clinical impact depends on reasoning and semantic understanding, not just massive repositories. Oracle’s new semantic database, he explained, is designed to close that gap and support both patient-facing tools and provider AI agents.
“Not only have we made the semantic database the underlying format for all of our clinical products going forward, but we’ve also made it available to APIs, partners, editors,” Mr. Sicilia said. “We even have competitors here in our demo grounds today, and I think it gives systems the opportunity to build healthcare agents themselves.”
Looking ahead
As the session wrapped, Dr. Gross emphasized that realizing AI’s potential will require intentional cross-industry coordination.
“Healthcare is incredibly broad, but across all of the verticals, the message and the goal is the same: we need to make sure that AI translates into real value,” he said. “It’s not just about the models. It’s about the deployment, and that means the whole ecosystem — from technologists, clinicians to regulators to institutions — have to work together for the future we’re building.”
