Imagine Manhattan. Now fill it with 500-storey skyscrapers and ten billion people who never sleep, all of them building things, talking, making more of themselves. Run that for eighty years without a pause.
That entire eighty-year city, David Friedberg told The Diary of a CEO, is roughly one second inside one cell. The people are proteins: about 10 billion of them per cell, in his figure, each one about a human's size on that scale, and the human body has around 10 trillion cells, all firing proteins and messages at each other continuously. "That's how complicated biology is," he said.
He reached for the analogy when the host asked whether AI would accelerate this field — by making new discoveries, or simply by running the tests faster. Friedberg's answer was that the complexity is the point. Nobody is going to write down a complete, deterministic account of what happens in that city. AI, he argued, lets researchers make useful predictions anyway.
The switches on the genes
The science Friedberg used as his worked example is epigenetics. Every cell in your body carries the same DNA, he explained; what makes an eye cell different from a heart cell is which genes are switched on and which are off. Genes that are on produce their proteins; genes that are off do not.
Those switches — chemical markers sitting on the DNA and on the histone proteins it wraps around — can drift. Alcohol, sunlight, cigarettes, burnt food: all of them can cut DNA, Friedberg said, and the cell is extremely good at repairing the damage, millions of times a second, all over the body. But each repair carries a chance that a switch ends up slightly out of place. The cell then starts making the wrong protein, or stops making the right one.
Do that for decades across enough cells and the tissue misbehaves. Skin wrinkles. The heart beats less well. The retina stops picking up light and passing on the signal as cleanly, and things blur. Hearing fades; thinking and reacting slow down. "That's what aging is," Friedberg said. "Aging is the core human disease." It arrives fairly evenly across the body, he added, because every cell is subject to the same statistical drift at roughly the same rate — which is why, at 46, he finds his hip, his eyes and his brain complaining at once.
A microdose of four proteins
In 2006 Shinya Yamanaka identified four proteins that, applied to an ordinary cell, push all of its switches back to the configuration of a stem cell — an embryonic state. From there, other proteins can move the switches forward again to make an eye cell, a skin cell, a heart cell.
The twist came later, in Friedberg's telling: instead of a large dose, researchers tried a microdose of the same four Yamanaka factors on a couple of cells. The switches moved, but not all the way back to a stem cell. An eye cell became a young eye cell; a skin cell a young skin cell. That partial move is what is now called epigenetic reprogramming.
It comes with an obvious failure mode, and Friedberg named it: reset a cell too far and it becomes an embryo and turns into cancer. The work of recent years, he said, has been finding the right dose of the right proteins in the right cells — enough to reverse age, not enough to erase the cell's identity. He cited mice living to a human equivalent of more than 100 years, wrinkles removed from monkey tissue, and dosing in the retina aimed at reversing blindness in people with a retinopathy. From there he made a strong claim: "The science is there. The foundation is there. We're just doing the engineering."
Where the eye work has actually reached
The eye is the furthest along, and the public record is narrower than the phrase "reverse blindness" suggests. Life Biosciences, whose platform uses controlled expression of three transcription factors — OCT4, SOX2 and KLF4 — to change gene-expression patterns in damaged retinal ganglion cells, announced on 9 June 2026 that the first participant had been dosed with ER-100 in a Phase 1 trial for optic neuropathies, including open-angle glaucoma and non-arteritic anterior ischemic optic neuropathy.
A Phase 1 trial's main purpose is to test whether a treatment is safe and tolerable; this one adds measures of visual function alongside that. The company attributed restored visual function to its preceding animal studies. The announcement reports a first dose in people, not a demonstrated restoration of human sight, and not established age reversal.
A million ideas, filtered on a computer
This is the gap Friedberg wants AI to close. Rather than understanding the Manhattan-sized cell, he described building models that predict which proteins might shift the epigenetic switches in a useful direction "without us having a deterministic understanding of what's going on."
The workflow he described runs in stages. Generate a million candidate ideas for a new protein. Test them in silico — on computers — to see how each would interact with everything else, whether it produces an effect like the known factors, and whether it might work better. Cut the million down to a short candidate list. Take that list into a laboratory, increasingly an automated one, and test on cells and in models to see whether anything actually reverses aging.
The alternative, he said, is a scientist having a good idea, waiting five weeks for the result, then having another good idea. To illustrate the difference he offered a number: a candidate whose probability of success over ten years might have been 1% arrives at something like 75%. He gave it as a rough comparison rather than a measured figure.
Asked whether he trains his own models or uses publicly available frontier ones, Friedberg said both. His teams build predictive models from their own data, use off-the-shelf life-sciences models, and also use general-purpose models. "So it's a great combination."
He closed the exchange on hiring rather than screening. The tools have not led him to cut jobs, he said: "It's not like AI got me to get rid of jobs. It's like the frontier has gotten so much wider" — new companies, new pathways, problems he says could never have been tackled before. The host's verdict on that was two words: "It is a persuasive argument."