Since the mid-1800s, U.K. life expectancy has consistently increased despite the occasional shocks from major global pandemics or wars. Since the early 2000s, there has been excellent progress in reducing rates of death attributable to cardiovascular diseases and certain cancers. However, these improvements slowed in the decade leading up to the COVID-19 pandemic and only last year did mortality rates return to the relatively low levels achieved in 2019, just before that pandemic.
With such significant progress and some of the more obvious gains in hindsight having been made (e.g. reduction in smoking prevalence), it is less clear what will drive significant further extensions in life expectancy. Furthermore, the NHS currently faces several significant challenges that appear to present a headwind to further gains.
In theory, the application of AI has the potential to bring about significant positive change in a great number of health-related areas that could lead to improvements in life expectancy.
Click on the sections below to explore some areas where AI may affect lifespan.
Being able to detect the potential onset of a disease before it takes effect can drastically improve health outcomes and is why certain existing screening programs are the norm in the U.K.. AI presents the possibility of reduced costs through automation, identifying new and less-invasive signatures for the early detection of diseases, and improved accuracy in pinpointing issues.
Some examples of disease detection include:
New drugs provide the hope of curing previously untreatable diseases or improving the effectiveness or patients' experiences of current treatments. AI could accelerate the discovery process significantly, potentially resulting in a ten-fold increase in the rate of drug discovery (which currently takes 5-10 years).
Some examples of accelerated drug discovery include:
Coined in the early 2000s, 'P4' medicine is Predictive, Preventive, Personalised and Participatory medicine – moving from one-size-fits-all medicine to an approach that is personalised to the individual in order to determine the optimal treatment. AI offers the ability to accelerate and enhance the art of the possible, predicting a patient's responses to medical interventions by using AI models that combine many large data sources (e.g. genomics, biometrics and historical medical records) to develop the treatment plan and medicine itself.
Some examples of enhancing precision (P4) medicine include:
The amount of surgery undertaken with robotic assistance has been growing fast in the last few years. Advances in robotics have enabled the performance of more intricate, complex operations less invasively and with enhanced precision. This can improve patient outcomes, with fewer complications, faster recoveries and less time in hospital, and can be particularly beneficial for elderly or frail patients who may not tolerate surgery as well. AI could expand the breadth of capabilities of robotic assistance in surgery environments, including the monitoring of and reaction to visual information and other data from patient monitoring systems.
Some examples of robot-assisted surgery include:
The combination of continual monitoring devices, AI and internet connectivity could enable improved management and long-term care in relation to a number of chronic diseases. For example, providing round-the-clock detection of subtle signs of deterioration in their condition would alert the individual, their caregivers or their medical professionals to act.
Better management of chronic diseases could lead to fewer (or the later onset of) severe complications, reducing co-morbidities and leading to longer, healthier lives.
Some examples of Chronic disease management include:
Smartphones, smartwatches and other wearables have been tracking all manner of metrics about us over the last decade and are now being enhanced with an array of AI features. These devices can provide targeted behavioural nudges that guide us towards improved physical and mental health as well as tailored plans to improve fitness and diet. The incorporation of large language models may lead to increased interaction with our devices due to the more natural human-like communication. Additionally, it offers a new channel for providing mental health support.
One of the pervasive AI opportunities across the economy is to improve efficiency. In the NHS, the potential areas that might benefit are broad, including:
Some examples of NHS efficiency and healthcare availability include:
Dario Amodei (CEO of Anthropic) suggested last October that we might see a century's worth of medical progress in just the next decade (which shares the same "10x" grounding in the comments from others in the technology and medical fields quoted earlier in this article).
The latest model for improvements in mortality rates would see life expectancy for a 65-year-old increase from around 23 years to around 30 years over the next 100 years. If we really saw that much medical progress in a decade, then the resulting 7-year increase to life expectancy over the decade could lead to a jump of c. 25% in the value of the liabilities of a 'typical' DB scheme. While such an extreme outcome might be theoretically possible, this would seem to require everything to go in the right direction rapidly and so perhaps in practice a smaller improvement is more probable.
Ascertaining what might happen in practice is fiendishly difficult and speculative. Additionally, improvements in areas might overlap, counteract or be offset in another area. Therefore, any estimates need to be taken with an unhealthy dose of salt. Nevertheless, the table below provides some high-level subjective illustrations of potential population-wide life expectancy increases, with a focus on scenarios in each area touched on above that could play out in the next decade.
| Area | Scenario | Illustration of plausible life expectancy increase* |
|---|---|---|
| Disease detection | Widespread roll out of broad multi-cancer screening program | 2 months |
| Accelerated drug discovery | Discovery of preventive Alzheimer's drug that slows the onset or cures the disease | Less than a month (slowing onset) to 3 months (cure) |
| Enhancing precision medicine | Doubling the effectiveness of cancer treatments through targeted cancer vaccines | 6 months |
| Robot-assisted surgery | Reducing fatality rates associated with surgery by a half | Less than 1 month |
| Chronic disease management | Reducing progression of diabetes and associated mortality rates | 6 months |
| Lifestyle improvements | General improvement in population lifestyle through better diet and exercise | 1 year |
| NHS efficiency and healthcare availability | Material reduction in deaths of population waiting for treatment | 4 months |
* Please note that these crude illustrations do not provide a view about the likelihood of such changes materialising nor cover all scenarios that could arise.
These plausible scenarios for the potential impact of AI show that, over the next decade, life expectancy improvements in the region of a couple of years of life expectancy are feasible, which could translate into an increase in the value of typical DB liabilities in the ballpark of 10%.
We are, of course, unlikely to see one-way traffic and AI could also present downsides to life expectancy prospects such as through job displacement, increases in sedentary lifestyles and exacerbated health inequalities. Additionally (though beyond the scope of this article), there are many factors, positive and negative, that will influence life expectancy beyond what is achieved through the application of AI.
In the light of the many areas where AI could have a positive impact on human lifespan and acknowledging the considerable momentum behind current investments in AI, it is hard not to be optimistic about the role AI might play in improving population-wide life expectancy and the nation's health over the next decade. Whether advancements are equivalent to a century's progress in a decade or just a substantial improvement, the potential benefits could well be significant.
Those in the pensions industry should keep a watching brief on developments in this area and consider carrying out scenario analysis to understand the potential risks for their schemes and appropriate risk-mitigation strategies.
