The Download: A Biological De-Aging Contest, and Why LLMs Don't Reason

Biological De-Aging Contest Draws 500 Competitors

Biological de-aging contest has officially begun, and roughly 500 people have signed up to compete. The premise is simple: over six months, participants will try to reverse their biological age using a range of different measures. There is even a leaderboard.

Jessica Hamzelou, who recently turned 40, is one of the competitors. She signed up this week for what she describes as an unusual competition, one that rewards people for getting younger. Her chronological age, as she puts it, only ticks forward. But the game here is not about birthdays. It is about biological age, a figure meant to offer a better way to measure the age-related health of our organs and bodies.

That distinction matters. Chronological age is fixed. Biological age is not. Or at least, that is the hope driving the contest. The question Hamzelou raises is whether it is even possible to measure whether someone is getting younger in any meaningful sense.

What Biological Age Actually Means

The concept of biological age has gained traction as researchers look for ways to capture how well the body is functioning rather than how long it has existed. Two people born on the same day can have very different health trajectories. One may have organs that function like those of a person a decade younger. The other may show signs of decline more typical of someone older.

Biological age attempts to quantify that gap. It is not a single measurement but a composite, drawing on various biomarkers and health indicators. The contest Hamzelou joined leans on this idea, asking participants to use multiple measures in pursuit of a lower biological age.

Here's the part the press release would skip. The science is still contested. Biological age is a useful concept, but pinning it down precisely is difficult, and that difficulty matters because different methods can produce different results depending on what they're measuring and how they're measuring it. So a person might look biologically younger on one measure and older on another.

The Leaderboard Problem

A leaderboard adds another layer. Competition can motivate. It can also distort. If participants know they are being ranked, they may work toward whatever the measurement rewards instead of genuine health improvements. That is a classic problem in any contest tied to a proxy metric.

Hamzelou does not shy away from the skepticism. She asks directly whether it is possible to measure whether someone is getting younger. The question hangs over the entire enterprise. Without a reliable measurement, the contest becomes a game of numbers that may not reflect real biological change.

But there's a catch. Even if the measurement is imperfect, the attempt itself may have value, because participants are likely to adopt healthier habits that stick with them long after the study ends. They may sleep better, exercise more, or change their diets. And those behaviors can improve health regardless of what the leaderboard says. That's what counts.

Why LLMs Do Not Reason

Thore Graepel doesn't hold back. In a separate opinion piece, he argues that large language models do not reason, a claim that lands with particular weight because Graepel serves as chair of machine learning at University College London and was, before that, a core member of the AlphaGo team at DeepMind, where the stakes of machine reasoning were never abstract. And he's not some outsider tossing stones. He says they don't reason. That's a serious charge from a serious insider.

Ten years ago, Graepel watched a program he helped build stun the world by beating Go champion Lee Sedol. AlphaGo won. It made a move so strange that some commentators thought it was a programming glitch, a moment that left viewers wondering whether the machine had broken or whether something far deeper was happening on the board. It was AlphaGo's powers of reasoning that made this creative choice. And these are powers that today's AI lacks.

That conviction drove him out. Graepel left Google DeepMind. He thinks we've got it wrong on machine reasoning, that a fresh approach is needed, and that this new approach should draw on AlphaGo's architecture, because the old ways don't get us where we need to go. But that's just his view.

The contrast is stark. AlphaGo operated in a closed system with clear rules and a defined objective. It could search through possibilities and evaluate outcomes. Large language models, by contrast, predict the next token in a sequence. They generate plausible text. They do not, in Graepel's view, reason through problems the way AlphaGo did.

What the Distinction Means

This debate isn't academic. It shapes how companies build AI systems, how regulators think about safety, and how users trust the output, and that's why it matters to everyone involved. If a model is reasoning, it can be held accountable for its logic. But if it's merely predicting, the failure modes are different.

Graepel's departure signals that at least some researchers think the current path has limits. The architecture that produced AlphaGo's creative move is not the architecture behind today's chatbots. That gap matters.

• Around 500 people signed up for the six-month biological de-aging contest.
• Jessica Hamzelou recently turned 40 and is competing.
• Thore Graepel left Google DeepMind to pursue a fresh approach to machine reasoning.
• AlphaGo beat Go champion Lee Sedol ten years ago.
• Graepel is chair of machine learning at University College London.

The Takeaway

Two stories, one theme. In both cases, measurement is the hard part. Biological age is difficult to pin down, and a leaderboard may reward the wrong things. Machine reasoning is difficult to define, and conflating it with fluent text generation may lead us astray.

Hamzelou's contest will produce a winner. Graepel's argument will produce more debate. Neither resolves the underlying question: how do we know if something is actually getting younger, or actually thinking? The answers are not in the leaderboard or the benchmark. They are in the architecture, the biology, and the honesty to admit what we cannot yet measure.

For now, 500 people will try to reverse their biological age. And one of the architects of AlphaGo will try to build a machine that reasons again. Both are worth watching.