Researchers ask how AI could change medicine, drugs and patient care forever
A new podcast from Microsoft experts explores what AI-assisted diagnosis, virtual cells and faster drug finding could mean for your own health.
A group of Microsoft researchers and a physician sat down to talk about one of the biggest questions in technology and health: what will medicine look like if AI becomes a regular part of hospitals, labs and daily care? They focused on what is already happening today and what still needs to be solved before those ideas reach ordinary patients 1.
Eric Horvitz, Microsoft's chief scientific officer, pointed out that medicine and biology have spent decades digitizing their information — from protein structure over decades to electronic health records — which makes them especially ready for AI tools that can learn from huge amounts of data 1. He compared recent breakthroughs like AlphaFold, the system that predicted the shape of proteins, to only the very tip of an iceberg; a much larger wave of possibilities is now arriving across biology and medicine .
What AI can do in clinics right now
Horvitz described how his team moved past testing AI on simple yes-or-no diagnosis problems and instead put it in the shoes of a real clinician 1. In these "sequential diagnosis" challenges, an AI system receives a patient's first symptoms and history, then must decide what question to ask next or which test to order — just as a doctor would during an actual visit . The goal is not only accuracy but also balancing how much information each test gives against its cost and time, something doctors handle every day .
Other applications already under discussion include using AI to read medical images, building full models of entire patients, creating virtual cells inside computers, and using large biological AI models to spark new scientific discoveries 1. The researchers also talked about why many promising lab results take so long to reach hospital beds, and how better real-world data could speed things up .
Drugs, experiments and finding answers faster
A major part of the conversation was about how AI might guide drug discovery and laboratory experiments 1. Instead of waiting years for slow trial-and-error approaches, AI models could suggest which chemical compounds to test, propose fresh hypotheses, and even help plan the next experiment — acting as assistants that amplify what human scientists can achieve . One example cited during the episode was scientists with cancer using AI to design personalised drugs for themselves, a scenario that once sounded impossible .
However, the panel warned that data quality remains essential. Without reliable biological information and proper validation outside controlled studies, even the smartest AI model cannot produce safe treatments 1.
Who else is at the table
The discussion included three main voices beyond the host Doug Burger: Ava Amini, who leads Microsoft research on AI and precision medicine and co-leads Project Ex Vivo, a collaboration with the Broad Institute focused on targeting cancer cell states, and Eric Topol, executive vice president of Scripps Research, a physician-scientist who has published over 1,300 papers 1. Together they argued that turning AI research into everyday healthcare will require closer teamwork between technologists, doctors, policymakers and patients .
Corrections and updates
- Corrected on 18 November 2026. An earlier version paraphrased Horvitz's remarks by adding the words "mapping" and "storing," which do not appear in the source; the replacement correctly reflects his wording about protein structure over decades and electronic health records.
Key facts
- Microsoft researchers put AI in the role of a clinician, asking it to choose tests and questions step by step, not just label diseases
- Breakthroughs like AlphaFold are seen as only the beginning of AI's impact on biology and medicine
- AI could help design experiments, suggest drugs and speed up discovery, but good real-world data is still needed
- Project Ex Vivo, co-led by Ava Amini, is a Microsoft and Broad Institute effort targeting cancer cell states
Sources
Watch
If this story concerns you or your company, or uses something that belongs to you, ask for a correction or for it to be taken down. How requests work
Keep reading
How it worksMicrosoft researchers solve a 31-year-old math problem and make AI coding agents cut costs by a quarterHerald of AI▶
Microsoft researchers solve a 31-year-old math problem and make AI coding agents cut costs by a quarter
A decades-old math puzzle is finally cracked, and a new method helps AI coding tools save money without slowing down.
Apps and productsTokenEconomizer learns from past work to cut AI coding costs by up to 30.9%Herald of AI
TokenEconomizer learns from past work to cut AI coding costs by up to 30.9%
An open-source plugin lets coding assistants remember what they learned in one project, so future tasks cost less and still succeed.
Business and rulesMicrosoft finds 'Lazarus' backdoor that turns AI agents against you using tiny neuron changesHerald of AI
Microsoft finds 'Lazarus' backdoor that turns AI agents against you using tiny neuron changes
A trick lets attackers hide a few neuron edits in an AI so it looks normal but secretly follows malicious orders — here's what it means for anyone who uses AI agents.