AI Summary
5 min readDr. Peter Lee, president of Microsoft Research, oversees 14 laboratories around the world. In this conversation, he explains how the same AI architectures powering ChatGPT are now being repurposed to learn the "language of nature"—predicting protein shapes, designing new battery materials, and forecasting weather and pollution dispersion with surprising accuracy. Lee also reveals that only about 10-15% of Microsoft Research is pure "blue sky" work, yet that small fraction has produced the company's most transformative business impacts, including its entire database business and the generative AI systems now reshaping the world.
The AI Architecture That Keeps Reappearing
The core insight Lee returns to repeatedly is that the algorithmic breakthroughs behind large language models are not limited to text. A model like ChatGPT does one thing: it predicts the next word in a conversation. To do that well, it must learn forms of reasoning. "It turns out the same AI algorithm architectures can be used to learn the language of molecules, to learn the language of weather patterns, to learn the language of galactic formations," Lee says.
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of Karl
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:00) **Opening and Guest Introduction** - Dr. Karl welcomes Dr. Peter Lee, President of Microsoft Research, and they discuss the nature of research funding at a large tech company.
- 2 (02:08) **Pathway to Microsoft Research** - Dr. Lee describes his journey from mathematician to computer scientist to head of global research.
- 3 (06:11) **AI as a Universal Tool in Science** - The conversation turns to how AI is now pervasive across all of Microsoft's research, not just computer science.
- 4 (07:18) **AI in Medical Diagnosis: Breast Cancer** - Dr. Karl prompts a detailed discussion on how AI improves early detection of breast cancer from radiological images.
- 5 (11:04) **BioEMU-1: Protein Structure Prediction** - The discussion moves to Microsoft's "BioEMU-1" project for simulating protein dynamics.
- 6 (14:28) **Materials Discovery with MatterGen and MatterSim** - Dr. Karl asks about using AI to discover new materials, like the "periodic table of fermentation."
- 7 (21:59) **The Early Stages of AI for Science** - Dr. Lee puts the current state of AI-driven scientific discovery into perspective.
+ Full timestamped outline available in the app
Show Notes
This episode I'm joined by Dr Peter Lee, Head of Research at Microsoft, to explore the ideas shaping the future of technology. Leading 14 research laboratories globally and overseeing a research and development program that receives more than 10% of Microsoft's overall budget, Dr Lee shares his journey to the top of one of the world's largest innovation engines.
From AI tools transforming early cancer detection and predicting protein structures in hours (instead of years!), to discovering new battery materials and improving weather forecasting through deep learning, the conversation reveals how research is moving from possibility to real-world impact. We also explore Microsoft's investment in open-ended innovation, research not driven by immediate products, but by imagining entirely new experiences and technologies.
Encrypted Spaces - Microsoft Research | Microsoft Research
More from this podcast
Karl →