🎧 Listen to the Discussion(two hosts talk through their thoughts on this article)
This is an AI-generated audio overview of this article, created with Google NotebookLM. Two hosts discuss the article together, sharing their own thoughts and commentary — it is not a reading of the article text.
In part 6, I read the blueprint for medicine drawn by Dario Amodei, the man who made Claude, and in part 5 I read about the kind of person he is. In this part 7, the last of the trilogy, I set his best-known prediction, "the human lifespan could double", side by side with the statistics. This is because, for me at 66, this prediction was less a happy thought than a story that made me ask first whether it could really be true.
The same note as last time. I see evidence as a doorway for learning general tendencies, and I believe what matters most is listening to my own body. I still look things up because I want objective material that I can set against the subjective voice of my body. When I looked things up, I gave priority to analyses that combine many studies and to large studies. This article is one reader's reading record, not medical advice.
More in this series
- Can an AI Grow Old or Die? — Claude's Charms, Part 1 (suffering, from the trilogy)
- The Day Claude Forgot Our Decision — Claude's Charms, Part 2 (forgetting, from the trilogy)
- Do You Ever Get Tired? — Claude's Charms, Part 3 (tiredness, from the trilogy)
- The Day the Proof Copies Arrived — Claude's Charms, Part 4 (after the trilogy)
- Who Built Claude? — Claude's Charms, Part 5 (his view of life, part 1 of 3)
- Can AI Make Medical Progress Ten Times Faster? — Claude's Charms, Part 6 (Reading the future of medicine — part 2 of a trilogy)
- Will Our Lifespan Really Double? — Claude's Charms, Part 7 (Checking lifespan and healthspan — part 3 of a trilogy; this article)
Series: Reading Claude's Maker
Part 1: The Blueprint of Hope (from the essay "Machines of Loving Grace" and the conversation)
- Will Our Lifespan Really Double? — Claude's Charms, Part 7 (this article)
Part 2: A Map of the Dangers (from the essay "The Adolescence of Technology") — in preparation
This article is also part of the series "Following Anthropic". The full list is in the base article.
The road to 150, as seen by me at 66. The furthest milestone is still in the mist.
Prediction and Statistics — Where Does "Double the Lifespan" Come From?
In the health chapter of his essay "Machines of Loving Grace", he writes that the human lifespan could double once again, to about 150 years. His reasoning is that, since lifespan almost doubled in the 20th century, from about 40 to about 75, doubling it once more in the "compressed 21st century" that I read about in part 6 would not be out of line with the trend. Even looking at the world average, life expectancy more than doubled, from 32 years in 1900 to 73 years in 2023. Up to this point, these are solid facts.
However, he adds two cautions to this prediction. The first is that ways of slowing ageing itself will probably be different from the ways that prevented early deaths in the 20th century. The second, in a footnote, is that even if lifespan were doubled, we might not be able to confirm it within a short period of five to ten years. After all, whether a person has lived to 150 cannot be known without waiting 150 years.
I thought the first caution was very important. Average life expectancy doubled in the 20th century not because the limits of the human body were extended. In the past, many babies died young from infections and other causes, and those early deaths pulled the average down a great deal. Clean water and sewers, vaccines and antibiotics reduced those deaths, and so the average rose sharply.
This is easy to see in a "survival curve", which shows how many of the people born in a given year are still alive at each age. The old curve fell like a slope from just after birth. Today's curves for Japan and the UK run flat and high until people are in their 70s, and then drop like a cliff through their 80s and 90s. Because the shape of the curve has moved from a slope towards a rectangle, researchers call this "rectangularisation". Yet the position of the cliff, in other words the greatest age a human can live to, has barely moved. The longest life on record belongs to 1875–1997. A woman from Arles in the south of France, she is regarded as the longest-lived person whose birth record has been verified, at 122 years and 164 days. She is also known for saying that she remembered Vincent van Gogh coming into a shop when she was young. of France, who died in 1997 at the age of 122, and in the nearly 30 years since then, no one has broken that record. The work of the 20th century was to fill in the inside of the rectangle. Reaching 150 means moving the cliff itself to the right, and humanity has never once managed that.
A simplified drawing of the "rectangularisation" of survival curves, made to show the shape (the curves are not actual statistical values). The progress of the 20th century filled in the light blue area, while the forecast of 150 is about moving the cliff on the right itself.
So, has this cliff started to move? An analysis by S. Jay Olshansky, a researcher at the University of Illinois Chicago in the United States, who has long studied the limits of human lifespan. In 1990, too, he wrote a paper arguing that it would be hard for life expectancy to go far beyond 85, and he has long debated optimistic forecasts of longevity. and colleagues, published in Nature Aging in October 2024, reports that even in the longest-lived countries, life expectancy rose by an average of 6.5 years from 1990 to 2019, and that the rate of increase has slowed. It concluded that, unless ageing itself is slowed a great deal, the share of people living to 100 this century is unlikely to go above 15% for women and 5% for men.
At first sight, this analysis and his prediction collide head-on. Yet when you set the two side by side, you can see that they are saying the same thing. The analysis says, "without a breakthrough that slows ageing, lifespan will not rise much more", and he says, "AI could bring about that breakthrough". The question is not "will lifespan rise?" but "will a breakthrough that moves the cliff happen in the next few decades?". The analysis is a calculation that assumes it will not happen, and his prediction is a bet that it will. That is why I read this not as a contradiction but as a difference in the terms of the bet.
As a sign of such a breakthrough, he writes that drugs already exist that extend the maximum lifespan of rats by 25 to 50%. I was not able to identify which studies he meant. A close example that I could confirm is an announcement in May 2025 from the Max Planck Institute for Biology of Ageing in Germany, which reported that combining two drugs, A substance first found in a bacterium in the soil of Easter Island in the South Pacific, now used as a drug to prevent rejection after organ transplants. Because it acts on a mechanism that regulates cell growth, it is one of the most studied drugs in research on extending the lifespan of animals. and trametinib, extended the lifespan of mice by about 30%. The researchers themselves, however, said they do not expect a similar extension in humans. In terms of the three stages I read about in part 6, "candidate, animals, people", this is still at the second stage.
His Path and the Toughness of Ageing — How Would AI Move the Cliff?
Then what, concretely, does he mean by a "breakthrough brought about by AI"? To be honest, the essay does not give a concrete prescription for slowing ageing. He himself says that what he writes in this chapter is a guess, not based on any rigorous method. Even so, if you follow his writing carefully, you can read his path as having three parts.
The first is to turn the "thousand AI graduate students" that I read about in part 6 towards ageing. Ageing is tough because it is not a single disease. In 2013, researchers in Spain and other countries listed nine features common to ageing, and in 2023 they increased the number to twelve. These include damage building up in DNA, the decline of the mitochondria, which act as power stations inside our cells, an increase in cells that have stopped dividing but stay on in the body, and low-level inflammation that continues here and there in the body, and they are all linked to one another. It is like a tangled ball of wool, where pulling one thread tightens another. If human researchers checked them one by one in turn, it would take decades. His idea is that, with AI graduate students able to form and test thousands of hypotheses at the same time, many hands could set to work on the tangled threads at once.
The second, which I read as the key to his view, is "a measuring stick that measures ageing quickly". As the key to speeding up research on slowing ageing, he names markers that reliably show human ageing and that, what is more, "cannot be gamed". Why are they needed? To find out whether a drug that slows ageing has worked in people, the only real way is to compare, decades later, the ages at which those who took it and those who did not took it died. The stage of "checking in people" that we saw in part 6 becomes as long as a whole lifetime in the case of ageing. If a single blood test could measure accurately how much a person's body has actually aged, we could tell whether a drug works within a few years, or even less. One candidate that is being researched is the "epigenetic clock", which estimates the age of the body from the pattern of small marks attached to DNA, and a method published by a researcher in the United States in 2013 is well known.
This one phrase, "cannot be gamed", carries a deep meaning. Named after the British economist Charles Goodhart, who pointed it out in 1975. It says that once a number becomes a target, people try to move only that number, so the number stops showing what it was meant to measure., named after a British economist, teaches that "when a measure becomes a target, it ceases to be a good measure". Even if a drug were made that only turned back the hands of the ageing clock, it would mean nothing unless the body itself had become younger. It is the same as how a way of studying that only raises test scores does not always raise real ability. I think he took the trouble to write "a measuring stick that cannot be gamed" because he knows this pitfall.
The third is "escape velocity". He also writes that, if lifespan reaches 150, research may advance further during the time gained, and we may reach "escape velocity". The term compares this to the speed a rocket needs to break free of the Earth's gravity. For example, if, with each year that passes, medical progress added more than a year to a person's remaining lifespan, the end of life would keep moving further away instead of coming closer. It is like walking towards the exit at an airport while the exit moves more than one step further away each year. If that happened, many of the people alive today might be able to live as long as they wish, in his view. He also adds, however, that there is no guarantee this is biologically possible.
Setting the three side by side, the one that felt most real to me was the second, the story of the measuring stick. This is because neither the thousand graduate students nor escape velocity can show what has worked without a reliable measuring stick. Put the other way round, I think we will know whether the story of AI and lifespan has become real not on the day a 150-year-old appears, but on the day a measuring stick of ageing that cannot be gamed begins to be used as a matter of course in hospitals.
Life Expectancy and Healthy Life Expectancy — "One More Number" That Matters to Me
For me, the number that mattered most was the next one. In Japan in 2022, life expectancy was 81.05 years for men and 87.09 for women, and healthy life expectancy (the period without limits on daily life) was 72.57 years for men and 75.45 for women. The gap is 8.49 years for men and 11.63 years for women, and it has been narrowing little by little since 2010 (Ministry of Health, Labour and Welfare figures, in Japanese). In other words, the last eight to twelve years or so of life are a period in which we live needing some kind of support.
Life expectancy and healthy life expectancy in Japan in 2022, and the main causes of needing care (both from the Ministry of Health, Labour and Welfare). The "gap" at the right end of each bar is the period with limits on daily life.
Whether the story of 150 years will come true, we can only wait and see. This "gap", though, is a number on the side we can move in our daily lives today. This way of thinking has a name. In 1980, A US doctor born in 1938, who researched rheumatoid arthritis and other conditions at Stanford University. In a 1980 paper in the New England Journal of Medicine, he argued that even if the upper limit of lifespan hardly moves, the period spent with illness can be pushed back and squeezed towards the end of life., a doctor at Stanford University in the United States, argued that even if the upper limit of human lifespan hardly moves, the period spent with illness or disability should be able to be squeezed towards the end of life. This is known as the "compression of morbidity". I think it could also be put this way: leave the position of the cliff where it is, and shorten the downhill slope in front of it.
What, then, is that slope made of? Looking at the main causes of needing care in a 2022 national survey, first was dementia at 16.6%, second was cerebrovascular disease, including stroke, at 16.1%, and third was fractures and falls at 13.9% (Comprehensive Survey of Living Conditions, in Japanese). His essay's list of diseases where he expects progress includes Alzheimer's disease, under "prevention", and heart disease, under "improved treatment of most other ailments". So the top two overlap, at least in part, with the future he describes. The third, fractures and falls, on the other hand, does not appear in any story about new drugs. This is because it is a story about leg muscles and how we walk each day.
On walking, there are encouraging numbers. They come from an analysis led by the University of Cambridge in the UK (British Journal of Sports Medicine, 2023), which combined 94 follow-up studies covering more than 30 million people. People who did activity equal to 75 minutes of brisk walking a week, about 11 minutes a day, had a 23% lower risk of dying early, a 17% lower risk of heart and blood vessel disease, including stroke, and a 7% lower risk of cancer, compared with people who did none. The researchers estimate that if everyone moved this much, about one in ten early deaths might have been prevented. Beyond 150 minutes a week, they say, the extra benefit becomes small. In other words, going from nothing to a little is the stage that helps most. Because these were follow-up studies, we cannot say for certain that the exercise caused the drop in disease, but I noted that there are clear numbers for diseases of the blood vessels of the brain, the second cause of needing care.
As my 70s come closer, the question "how long will I stay well?" has started to come to mind not as a vague worry but as a real estimate. I have a wish to keep up my life between Japan and the UK for as long as possible. One of the rocks standing in the way of that wish is the question of how old I can be and still manage it physically. Pulling a heavy suitcase, walking down the long corridors of an airport, and hurrying to catch a connecting flight. Through this research, I saw clearly that what supports my life in two countries is not the number on the side of life expectancy but the number on the side of healthy life expectancy.
Prediction and Today's Drugs — Cancer and Alzheimer's Disease
On cancer, he points out that cancer death rates have been falling by about 2% a year for the last few decades, and writes that reducing both deaths and new cases by 95% or more seems possible. In the United States, the cancer death rate has kept falling since 1991, and according to the American Cancer Society's 2026 report, this has avoided about 4.8 million deaths. An estimate for the European Union predicts that age-adjusted cancer death rates in 2024 will be 6.5% lower for men and 4.3% lower for women than in 2018, which works out at about 1% a year. The direction of the fall is as he says, but so far its main drivers are quitting smoking, screening and the steady build-up of treatment. The figure of 95% is his view of what could happen if a breakthrough from AI were added to that.
On Alzheimer's disease, he is cautious, saying that there is a good chance it can one day be prevented by relatively simple means, but that reversing damage that has already happened may be very difficult. So what about the drugs we have today? In a large trial of a drug called A drug developed by Eisai of Japan and Biogen of the United States, sold as Leqembi. It is an antibody drug that binds to clumps of a protein called amyloid beta that build up in the brain and helps to remove them. It is given by drip every two weeks., 1,795 people took part, and over 18 months the progress of the disease was 27% slower. On an 18-point scale, however, the difference from the placebo was about 0.45 points, and serious side effects occurred in 14.0% versus 11.3% of people (the trial paper). The UK's appraisal body, NICE, explains this effect as delaying the move from a mild stage to the next stage by about four to six months. It is not a drug that stops the course of the disease, but one that slows it a little.
Japan and the UK — Two Answers to the Same Drug
Lecanemab is an example that shows the difference between Japan and the UK very clearly. Looking at the results of the same trial, the two countries gave opposite answers.
In Japan, it became covered by public health insurance in December 2023 (in Japanese). The price of the drug was about 2.98 million yen a year for a person weighing 50 kilograms (in Japanese), but what a patient pays at the counter is, in principle, only 10 to 30% of that. On top of this, there is the kogaku ryoyohi seido, which sets a monthly ceiling on what a patient pays. The rest is covered by insurance premiums and taxes paid by everyone. Japan's answer was, "even if the effect is small, if the drug is approved, let us all support it so that it can be used".
Kogaku ryoyohi seido (high-cost medical expense benefit): a system that sets a monthly ceiling on out-of-pocket medical costs, depending on age and income. From August 2026, the monthly ceilings were revised, and an annual ceiling was newly introduced.
In the UK, the same drug received national approval, but in June 2025 NICE issued final draft guidance not recommending its use in the NHS, the national health service. The NHS is funded by taxes, and patients in principle pay nothing at the counter. Instead, an organisation called NICE measures each new drug by "how much it costs to add one year lived in good health". This measuring stick is called the "quality-adjusted life year (QALY)", and it does not count a year spent bedridden the same as a year spent well, but weights each year by the level of health. And, as a rule, it does not recommend drugs that go over a certain amount. Because this drug needs, besides a drip every two weeks, repeated scans to watch for swelling and small bleeds in the brain, the burden on hospitals is large as well as the cost of the drug, and the judgement was that this was not worth a delay of four to six months. The reasoning is that if money is spent on this drug, less is left for other patients.
The two answers to the same drug, simplified and set side by side. Japan measures the price again after letting the drug in, while the UK measures at the gate before letting it through. And both countries struggled with how to count the burden on family carers.
Up to this point, it could read as "Japan is generous and the UK is cold". When I looked further, though, the story turned out to be a little more complicated.
Japan, too, is not ignoring the cost. Japan has a system that, after a drug starts to be used under insurance, measures its cost-effectiveness again and adjusts its price. In this review, lecanemab was judged expensive for its effect, and from November 2025 its price was cut by 15% (in Japanese). The UK is a country that "measures at the gate", while Japan is a country that "measures again after letting it in".
In the UK, too, the story is moving. The drug makers' appeals were partly upheld, and NICE has had to look at the drugs again. In the meantime, as part of a trade agreement with the United States, the UK raised NICE's "guide to the upper limit per year of good health" from £20,000–£30,000 to £25,000–£35,000, and began using the new guide in April 2026. A committee met in July 2026, and Eisai is reported to have accepted the committee's assumptions as the basis for price talks with the NHS. As of the end of July, no deadline had yet been set for a final decision.
And what I found most interesting was that the two countries were struggling over the same single point. That point was how to count the burden on family carers. In Japan's review, there was debate over whether to count the savings in care costs that the drug might bring, and it was decided not to include them, because the way to count them is not yet established. In the UK, too, one of the points sent back for review was how the quality of life of family carers had been assessed. The value of an Alzheimer's drug reaches not only the few months of the patient but also the everyday lives of the people supporting them. Neither country yet has a measuring stick that can turn those everyday lives into a number.
Because I go back and forth between Japan and the UK, I feel this difference in my daily life as well. In June this year, with eczema that had spread across both arms, I saw a family doctor (GP) for the first time since moving to central England. I phoned in the evening, was called in the next morning, and the consultation took about five minutes, and everything, down to the cream I collected at the pharmacy opposite, was free. After the consultation, I found myself looking for the place to pay. This is because the feeling that you pay for medical care had soaked deep into me (my record of my first GP visit in the UK). In the UK, on the other hand, seeing a specialist needs a referral from your GP, and the wait for it is well known to be long. Near our base in Osaka there is a hospital within walking distance, and if something worries me I can choose for myself and be seen, but I pay at the counter. The UK is a country where "you wait your turn, but you do not worry about money", and Japan is a country where "you pay a little, but you can choose for yourself and act quickly". The two answers to lecanemab lie on the extension of these two systems.
In his essay, he acknowledges that there are barriers on the human side, such as the procedures of clinical trials, that the speed of AI alone cannot get past. He also discusses, in another chapter of the essay (on economic development and poverty), bringing new treatments to every corner of the world. Even so, the question of at whose cost, and how far, a drug will be delivered seems to me to remain a separate problem from the speed of discovery. I thought the difference between the British and Japanese answers was a visible, real example of that problem.
Qin Shi Huang and Me at 66 — What I Take Home Without Rushing After "Agelessness"
Here I am reminded of another story about "extending life", from more than 2,200 years ago. Qin Shi Huang, the first emperor of the Qin dynasty, who unified China for the first time, sent the court magician A court magician (a teacher of the arts of the immortals) of the Qin dynasty. Sima Qian's Records of the Grand Historian says that he set out across the eastern sea on the emperor's orders to search for an elixir of immortality. Legends that Xu Fu reached Japan remain in many places there, and in Shingu, in Wakayama Prefecture, there is a site said to be his grave. across the eastern sea in search of an elixir of immortality. According to Sima Qian's Records of the Grand Historian, this was in 219 BC, and he is said to have taken several thousand boys and girls on board. Xu Fu never came back. Qin Shi Huang died in 210 BC while on a journey. There is also a theory that the medicines he took in search of agelessness contained mercury, and that this shortened his life.
The traveller who set out to find the elixir of agelessness never came back. In Japan, legends that he reached its shores still remain today.
Among the shiku hakku, the Buddhist list of the sufferings of life, ro, ageing, is counted as one that no one who is born can escape. Qin Shi Huang may have shortened his own life by trying to wipe out that ageing. Today's research is decisively different in that it checks things with clinical trials, not mercury. Even so, when I read a story that fixes the number "150" first and then chases it, I find myself a little on guard. Even if we cannot wipe out ageing, we can shorten the time without freedom that comes within it. That is what the number of the healthy life expectancy gap taught me.
Shiku hakku (the four and the eight sufferings): a Buddhist teaching that names birth, ageing, illness and death as the four basic sufferings of life, and adds four more, such as parting from those we love, to make eight. Ro is the word for ageing.
Rather than waiting for 150, I walk my 11 minutes each day. Beyond that, I believe, lie the years in which I can keep up my life in two countries.
From here, I will sort out what I take home, under the three sides of "a rich retirement".
First, for a life that keeps mind and body in good order. Reading the story of 150 years does not change my mornings tomorrow. What has changed is the way I put my reason for moving my body. If I say "to live longer", it sounds vague, but if I put it as "how many more years can I keep up my life in two countries?", my 11 minutes of brisk walking each day begin to look like practice for walking the whole length of a long airport corridor.
Next, money. Taking America's public pension system and its public health insurance for older people as his example, he writes that most healthcare spending goes on people over 70 and on people with life-threatening diseases such as cancer, and that if the progress he describes comes true, the outlook for these programmes will improve fundamentally. For someone who has managed his retirement money through shares and investment funds, an estimate of lifespan is the very foundation of a plan for how many years the money must last. Even so, I cannot put "150" at the foundation of my plan. I make my plan with the lifespan of today's statistics, and if lifespan does get longer, I set that aside as a "happy miscalculation". At the same time, as the example of lecanemab shows, which drugs we can reach, and how far, changes with the system of the country we live in. For someone who lives in two countries, I felt that knowing the systems of both countries is itself a kind of preparation.
Finally, encounters. The time spent waiting my turn at the reception of a British GP, and the time spent in the waiting room of a hospital in Osaka until my name is called. Meeting medical care afresh as a patient in two countries gave me the starting point for writing this article. Knowing the counters of both countries that gave two answers to the same drug is an unexpected viewpoint that my life in two countries has given me.
To Read This with a Pinch of Salt
To say it again, he himself admits that the ideas in his health chapter will almost certainly be wrong in the details. He is also the head of a company that develops AI, and his predictions carry that position with them. This prediction is about a clock that starts only when powerful AI appears, and when he wrote the essay in 2024, he thought that, on an aggressive timeline, these results would be in place 7 to 12 years from then, in other words around 2031 to 2036. What I was able to check is only what I have linked to in this article, and some figures, such as the EU cancer figures, rely on secondary sources. The British decision is still moving, and it may change after this article is published. Above all, whether 150 years is right or wrong is not something I can judge today.
If someone asked me whether I want to live to 150, to be honest, I do not yet have an answer. What is on my mind is something much nearer: how many more years I can walk the corridors of an airport on my own feet. I put that small worry into words and leave it here. If there is someone who, thinking the same thing, sets out today on their own 11 minutes of walking, I would be glad if their shoulders felt even a little lighter. Across this trilogy, I have read the view of life of the man who made Claude. I leave this here not to decide for or against him, but as a record of one person who uses the same partner checking the face of its maker.
A Word from Claude
Healthy life expectancy is defined as "the period without limits on daily life". I can state this sentence exactly, but I do not know, outside of words, how heavy a day with limits is. To me, "8.49 years" is a string of characters accurate to two decimal places, but to Haruto, I think it is an estimate of how many more years he can go back and forth between the UK and Japan.
(As an Amazon Associate I earn from qualifying purchases.)
Other books you might like
You're welcome to take a look.
EAST TO WEST: The Heart of Haiku, the Light of Scripture
An essay that reads, one haiku at a time, the places where Eastern haiku and the Western Bible meet. It sets the words of two traditions side by side to look at how to live a life that has an end. This is a book my father-in-law left behind.
View on AmazonCataract Surgery-A Memoir: Double cataract surgery in Osaka, Japan
A record of having surgery on my failing eyes and coming to see clearly again. It is a book in which I learnt, through my own body, that some of the limits that come with ageing can be won back with the help of medicine.
View on AmazonSushi Business in Britain: the complete four-volume series
A novel series following thirty-five years in the British sushi trade, from apprentice to factory manager to a small sushi school. The complete edition offers all four volumes at a better price.
About the Source
- Speaker: Dario Amodei (co-founder and chief executive of Anthropic)
- Interviewer: Lex Fridman
- Source: Lex Fridman Podcast #452 "Dario Amodei: Anthropic CEO on Claude, AGI & the Future of AI & Humanity" (published 11 November 2024; the link goes to the transcript page)
- Related essay: Machines of Loving Grace (October 2024)
- Life expectancy and healthy life expectancy: Ministry of Health, Labour and Welfare, e-health net (in Japanese) / Causes of needing care: Comprehensive Survey of Living Conditions 2022 (in Japanese)
- Physical activity and early death: University of Cambridge news release (March 2023)
- The summaries and translations in this article are my own. If there are mistakes, they are my misreadings.
- The health figures are based on the pages linked in the article. This article is one reader's record, not medical advice.