Mass General Brigham’s AI intake tool gets patients to a physician the same day. Its CFO is fine losing money on a third of its AI spending. And not one executive Becker’s spoke with over two days at the Somerville, Mass.-based health system’s innovation forum would call any of it fully autonomous — yet. Here’s what else stood out.
1. Mass General Brigham’s AI intake agent gets patients seen the same day, 77% of the time.
Mass General Brigham COO Ron Walls, MD, described an AI agent embedded in the health system’s MyChart patient portal that conducts intake for patients seeking primary care the way an intake nurse would: gathering symptoms, assessing urgency and routing the patient to a live visit, a virtual visit or urgent care. The tool, called MGB Care Connect, launched about a year ago and saw 36,000 patients in its first year; the system now sees 150 patients a day through it. “77% of the patients who access this see the doctor the day they accessed it, which is profoundly different from our usual primary care model,” Dr. Walls said during a panel discussion at the World Medical Innovation Forum, which took place Sept. 22-23 at the Westin Boston Seaport District. By the time a physician sees the patient, the AI has already drafted a note summarizing the visit, pulling in relevant history from the medical record.
2. No one is calling AI ready to act fully on its own in clinical care.
Across separate conversations, health system and government leaders converged on the same boundary. “So right now we’re not allowing autonomous AI across different platforms,” Mass General Brigham Chief Information and Digital Officer Jane Moran told Becker’s. Rebecca Mishuris, MD, the system’s chief health information officer, put it more bluntly at a panel: “Truly autonomous, like no human makes a decision or helps the patient make a decision — I’m not sure we’re there for anything yet.” Haider Warraich, MD, a cardiologist and program leader at the federal Advanced Research Projects Agency for Health, said at the same session that’s why his agency is investing heavily in agentic AI research.
3. “Pilot” is becoming a dirty word.
Dr. Walls said Mass General Brigham has deliberately avoided AI pilots. “A pilot … only tells you the technology can work in your system,” he said. “What you really need to know is: Is there a problem that can be solved, and can technology aid in that solution?” Christine Schuster, BSN, RN, president and CEO of Concord, Mass.-based Emerson Health, described a similar calculus born of tighter margins, telling Becker’s her system doesn’t have the bandwidth “to do a lot of ‘oh, let me try this out and see if it works, and it doesn’t.'” Kyle Bryant, who leads healthcare and life sciences work at Ode, an Anthropic-backed AI implementation firm, said at a panel that he coaches clients to “pick a thin slice, a workflow, a segment, something low risk to start with” rather than run open-ended pilots.
4. Mass General Brigham’s CFO won’t chase a return on a third of its AI spending — on purpose.
Niyum Gandhi, Mass General Brigham’s CFO and treasurer, told Becker’s the system sorts its AI investment into three categories, and only one is meant to generate a financial return. Spending on research and pure clinical-outcomes work, he said, “is always going to be a net investment because we aren’t looking for an ROI there.”
5. Revenue cycle has a problem: AI fighting AI.
As payers automate prior-authorization denials, some providers are beginning to fight back with AI of their own. “We have the bot wars,” Colin McHugh, president and CEO of Nashua-based Southern New Hampshire Health, told Becker’s, describing a shift on the payer side from human-reviewed denials to algorithm-driven ones. Ms. Schuster described the toll on physicians: Fighting a denial can mean a surgeon personally walking a payer through the clinical risk of, for example, discharging a bariatric surgery patient the same day.
6. A decade of “AI will replace radiologists” predictions still hasn’t come true.
Terrence Chen, MD, CEO of United Imaging Intelligence America, said at a session that radiology was AI’s natural proving ground because its data was already digital — but the field has defied a decade of displacement predictions. “Over a decade after people started to predict AI will replace radiologists, it certainly hasn’t happened yet,” he said, “and in many settings, radiologists don’t even feel that AI is making them much less busy.” Mass General Brigham Chief Innovation Officer Chris Coburn told Becker’s: “There’s a need for more radiologists, not less,” he said, “because radiology is more than just reading those images.”
7. Patients are skipping hospital websites for AI chat — and marketing teams are tracking it, not panicking.
Mark Bohen, Mass General Brigham’s chief marketing and communications officer, told Becker’s the system has seen fewer clicks to its newly consolidated website as patients research care through AI search tools and chatbots instead — a trend he said is showing up across many industries, not just healthcare. But he said the system hasn’t seen a drop in qualified leads, the patients who actually convert to appointments, because it tracks each AI platform separately and optimizes for the moment patients are ready to book care rather than competing as a medical encyclopedia.
8. Mass General Brigham Ventures’ spinout pipeline is paying off.
Roger Kitterman, managing partner at Mass General Brigham Ventures, pointed Becker’s to some of his department’s most successful investments: A hyperparathyroidism drug that originated from Mass General research and was spun out into a company called Amolyt Pharma was acquired by AstraZeneca for roughly $800 million upfront, with up to $1 billion more tied to milestones, he said. Other spinouts include CodaMetrix, an AI medical-coding company, and InStride Health, a virtual anxiety-treatment program that began at Belmont, Mass.-based McLean Hospital and is now covered by insurance in roughly two-thirds of states.
9. In health equity, the technology usually isn’t the bottleneck — trust is.
Elsie Taveras, MD, Mass General Brigham’s chief community health and health equity officer, told Becker’s the biggest barrier to getting new tools and treatments to underserved communities is rarely the innovation itself. She cited research showing it can take 17 years for a new therapeutic or technology to reach the communities that need it most. Without “the human infrastructure to match” new technology, she said, there’s no “vehicle for trusting relationships” to carry it to the people who need it. Her office is preparing to roll out ThriveLink, an AI-based telephonic tool that connects patients with food insecurity to services, systemwide.
10. The federal government is funding autonomous AI for heart failure care — but is requiring a human supervisory layer first.
Dr. Warraich said ARPA-H is funding teams — including ones led by Stanford University with OpenAI and by the University of Washington — to build both the clinical AI tools and a separate supervisory layer to monitor them after deployment. “I do agree that not everything needs to be monitored in real time,” he said, “but if you’re going to build a technology that can take such actions, you need a monitoring system that can keep pace with that.”
