The human cell is wildly complex. Can AI decode it? | Silvana Konermann
6/13/202620 min
Silvana Konermann and the team at Arc Institute are trying to crack one of science's most difficult problems: why complex diseases like Alzheimer's and cancer remain so stubbornly unsolvable, even as research advances. Her solution is a universal “virtual cell” — an AI model trained on a billion biological experiments that can read the language of human cells, predict what's going wrong and reveal how to fix it. In conversation with TED’s Chris Anderson, Konermann explores how this work could fundamentally change the way we discover drugs and treat disease. (This ambitious idea is part of The Audacious Project, TED’s initiative to inspire and fund global change.)
Hosted on Acast. See acast.com/privacy for more information.
Clips
Transcript preview
First 90 secondsElise Hu· Host0:00
[gentle music] You're listening to TED Talks Daily, where we bring you new ideas to spark your curiosity every day. I'm your host, Elise Hu. I'll be honest, before this conversation, the term virtual cell wasn't something that existed in my vocabulary. Turns out it's a real thing and could have major implications for some of the most complex diseases we know. Alzheimer's, for example, has stumped the medical field for decades because each patient's biology is uniquely tangled. But bioengineer and neuroscientist Silvana Konermann, who is a 2025 Audacious Project grant recipient, thinks that artificial intelligence holds the key to finally help us untangle it.
Silvana Konermann· Guest0:45
We've just seen over, I would say really the last two years, that it's getting real. I think that within four years, five years, we will be able to have these models, um, that are accurate enough to be useful and then it's a totally different way of doing biology.
Elise Hu· Host1:01
Silvana works at the Arc Institute, where she and her team are using single-cell sequencing or CRISPR, as well as AI, to run a billion physical cellular experiments. In other words, they're training a model that can speak the language of cells similar to the way large language models learn to speak ours. The goal? A universal virtual cell that tells researchers exactly which interventions could turn a real diseased cell back into a healthy one. It would