David Baker brings made-from-scratch proteins to The Royal Society
His Croonian Prize Lecture moves generative AI beyond images, into protein switches, plant defences and new enzymes.
Why this story? The channel's weekly summary: what went up this week, what is drawing views now, and where the channel stands among similar channels.
Advances in computational protein design | The Royal Society
Things we can learn from the video
Press to jump
- 0:00Baker introduces protein design as a way to programme matter at the atomic scale.
- 1:27RFdiffusion uses a denoising approach inspired by image generators to design protein structures from random noise.
- 7:51Conditional protein switches can activate or deactivate in response to specific signals, including tumour markers.
- 12:41Designed ligand-gated channels offer a way to trigger calcium flow for neurobiology research.
- 13:51A designed plant-defence system initiates targeted cell death only when a pathogen is detected.
- 14:55Baker presents newly designed enzymes that break chemical bonds, including those in organophosphates and peptides.
"I'm always the token person doing something good with AI," David Baker jokes during his Royal Society lecture. His examples give that line substance: proteins designed not merely to take a particular shape, but to perform a particular job.
In "Advances in computational protein design | The Royal Society", Baker explains how diffusion methods, inspired by image-generation tools, can build protein structures from random noise. Rather than modifying proteins found in nature, his laboratory designs new ones from scratch, produces the genes encoding them and tests the results experimentally. Baker shared the 2024 Nobel Prize in Chemistry for his work on computational protein design.
The intriguing step is from proteins that bind to something to proteins that respond to conditions. He presents switches intended to activate in the presence of particular signals, alongside a plant-defence system that triggers targeted cell death when it detects a pathogen. These are molecular designs with instructions attached.
At roughly 21,000 views, the lecture is the most watched of The Royal Society's five new uploads and its busiest video for daily viewing. AI also appears in the shorter "Is AI sentient? 🤖#ai #tech", drawn from the Michael Faraday Lecture 2025.
Across the channel, daily viewing is running below its usual level, though the subscriber total grew by about a thousand over the week. The programme itself remains broad: vaccine safety, Johan Håstad's account of complexity theory, and a fresh physics talk from Kymani Armstrong-Williams. That last upload, "This shouldn't be possible! | The Royal Society", connects the unresolved relationship between gravity and quantum mechanics to something as ordinary as jumping.
Published on Trari
Where next?
The videos drawing views on The Royal Society right now
- Channel
- The Royal Society
- Subscribers
- 257,000 ▲ 1,000 this week
- Channel views, last 7 days
- 110,910
- Videos published this week
- 5
- Video published
- November 3, 2026
- Video views so far
- 21,422
- Views on its first day
- 6,960
The numbers
Here is how The Royal Society's uploads and audience moved over the week.
💡 Test your knowledge
Which image-generation tool does Baker mention when explaining the inspiration for protein diffusion models?
Baker cites DALL-E and Stable Diffusion when explaining how denoising methods can generate protein structures.





