Read Reflections with pictures and graphs (pdf)
Few things can cause a macroeconomist to worry as much as when reading future forecasts of demographic developments. In large parts of the world, far too few children are now born while we are living longer. We have had to get used to the idea that an ageing population will become an increasingly heavy burden on society – where fewer young people will have to support more and more old people. How big the burden for young people will be depends largely on how far into their years the elderly can continue to work, but at least as important is how long they can manage on their own without having to be taken care of by society and healthcare. The key to both of these things lies in how long we can stay healthy. Like so much else in society, health and longevity is an area that many believe will be revolutionized by AI.
Biological AI revolution
So how likely is it that Demis Hassabis will be right, and if so, how could it happen when AI finds cures for most diseases? Well, these questions hardly belong to an economist's domain, but based on what is said by those who know more about the connection between AI and medicine, the short answer could be summarized as that fast digital simulation replaces slow human trial-and-error in the search for new, better medicines. Traditional medical research is like looking for a needle in a haystack. It takes an average of more than a decade to develop a new drug and the costs often amount to hundreds of millions of dollars. AI has the potential to fundamentally change this process by moving biology from labs and test tubes to computers and data centers. Instead of slowly developing medicines, we are moving towards letting AI design them. With the help of AI, biology can in practice become a software problem. A breakthrough that illustrates the power of this technology is the AI AlphaFold, which can predict how proteins – the building blocks of life – are folded three-dimensionally, which is crucial for how they function and interact with other substances. Before AlphaFold, it had taken human researchers around the world 60 years to map the appearance of 0.1 per cent of the more than 200 million proteins that exist. AlphaFold mapped the remaining 99.9 per cent in less than a year.
One of the absolute biggest bottlenecks in medical research is the time required – testing drug candidates first on cells, then on animals and finally on humans takes a very long time. With AI, you can instead build advanced "digital twins" of human cells and organs that AI can then run millions of simulations on. We can test how a new medical molecule interacts with human biochemistry directly in the computer before the molecule has even been manufactured in reality. This can cut the development time of a drug to as little as a few months. It is this kind of exponential development that can be found in many areas where AI is making its entrance, and it is what suggests that Hassabis's ten-year timeline does not have to be science fiction.
Are we moving towards eternal life or just fewer ailments?
If we hypothetically assume that Demis Hassabis is right, what kind of development are we facing? Does this even mean that "everlasting life" is within reach? One such scenario is called Longevity Escape Velocity (LEV). It simply means that medical science succeeds in extending our remaining life span faster than time passes. If every year research manages to add more than 12 months to your life expectancy, then you have in theory reached a point where you can live as long as you want. However, most gerontologists and researchers believe that the dream of "eternal life" is far away. Instead, it is more likely that we will continue to age, but that we will be less ill and frail in the last 15-20 years.
Won't the earth get full?
Although the demographic debate today is mainly about falling birth rates, the idea of radical life extension often raises a deep-rooted concern about acute overpopulation effects and that we would eventually be forced to stop having children. But if we actually look at the numbers, the picture becomes significantly less dramatic.
Today, the world's population amounts to about 8.2 billion. Every year, about 135 million new people are born, while about 60 million die. Today, the world's population is growing by about 75 million people per year. If we assume the most extreme scenario – that almost no one dies of disease while childbearing continues at today's level, the annual population growth would increase from 75 million to 135 million. After ten years, the world's population would land at just over 9.5 billion people. This is a faster increase than the UN's current forecast, which predicts around 8.9 billion by the mid-2030s. But it is far from a planetary collapse. In any case, we will have plenty of time to think about how we should deal with the population issue in the long run.
"The Longevity Dividend"
But what will be the financial consequences? In traditional economic theory, older people are seen as a dependent part of society because they no longer work and at the same time consume most of society's healthcare resources. But in a world without diseases, the line between "working age" and "retired" is erased. Some researchers, such as economics professor Andrew J. Scott at London Business School, have shown that there are enormous economic benefits from an extended health span – which he calls a Longevity Dividend. Scott's calculations suggest that if we extend healthy life expectancy by just a single year, it is worth about 3-4 per cent of GDP annually for a country like the United States, through a combination of reduced healthcare costs and increased productivity.
Experienced, highly educated 75-year-olds can continue to work or run businesses with the energy and mental capacity of their younger days. The health sector, which today swallows up large parts of a country's tax revenues (often over 10 per cent of GDP in the Western world), can be redistributed to other areas. In fact, one of the greatest threats to the modern welfare state today is that we live too long with poor health. Perhaps AI could shorten that time significantly.
Will technology only belong to the richest?
A common dystopian sci-fi trope is that only tech billionaires will be able to afford the new revolutionary AI drugs. But there are strong reasons why this is unlikely to be the case: once a treatment has been developed, the expensive part – the research and discovery – is already taken. The production itself is cheap and the marginal cost of manufacturing the molecules is usually low, and for governments and healthcare systems, it is also beneficial to distribute the technology widely. The cost of providing the population with AI-designed preventive medications is likely a fraction of what it costs to provide years of chronic dementia care, cancer treatments, or other expensive care.
History also shows that medical advances quickly become mass products. At the same time, one should not close one's eyes to another historical pattern: new medical technology has over time almost always driven up total healthcare costs as we find new things to treat, enabling us to live longer in poor health. For AI to break this trend, the technology must succeed in preventing disease altogether, rather than just prolonging treatment pathways.
Who can get on the train?
If the 10-year timetable holds, which of us alive today will benefit from this? Today's young people (generation Z and Alpha) may even be the first who may actually be faced with the question of choosing how long they want to live. But even for those of us who are approaching retirement age today, there is hope. If AI breakthroughs in biotech accelerate in the next decade, it will be enough to stay on track until 2036 to start benefiting from the first waves of regenerative medicine. Then you can buy yourself enough time to be on the next wave of innovations – and thus ride the "escape velocity" wave.
It is far from certain that AI will succeed in generating the health benefits that Hassabis predicts, at least not within such a tight time perspective. But if that happens, we should not underestimate the institutional challenges that lie ahead. Pension systems must be rebuilt from the ground up. Who wants to retire at 65 if you potentially have 50 healthy years left? At the same time, there is of course also the potential and challenges of AI's entry into the labor market, but it is too big a discussion to deal with here.
In conclusion, it is fair to state once again that AI is an invention that has the potential to fundamentally reshape the entire society and completely change the conditions for all of us. One could even argue that most of the other things that are happening right now are to be regarded as noise in comparison to the effects of the AI revolution.