What medical breakthroughs could mean for insurance and retirement by 2040
AI’s biggest impact on insurance and retirement may have little to do with automating insurance
Most discussion about artificial intelligence in insurance and wealth management focuses on what AI will do inside our industry.
Automated underwriting. Faster claims. Personalised advice. Fraud detection. Administration efficiencies. Better customer service.
All important.
But perhaps we are looking in the wrong place.
AI’s much greater long-term impact may come from what it does outside financial services — particularly to medicine.
AI is accelerating diagnostics, drug discovery and personalised medicine at the same time as advances are being made in genetics, epigenetics, cell therapies, regenerative medicine, robotic surgery, bionics and bioengineering.
By 2040, these technologies may start changing some of the fundamental risks our industries have spent decades learning how to price:
When do we die?
How long do we remain sick or disabled?
What conditions can be treated or reversed?
How long do we remain healthy and economically active?
And ultimately:
What does retirement actually mean?
That should matter today to anyone designing a life insurance, superannuation or retirement-income product that may still be operating decades from now.
From predicting disease to changing it
I first wrote about genetics and insurance more than a decade ago. My argument then was not that insurers should focus primarily on the potential for genetic information to create anti-selection. It was almost the opposite: insurers should encourage people to better understand their genetic predispositions, because that knowledge could allow them to change their lifestyles, undertake earlier screening and take preventative action to reduce the likelihood or severity of future disease. Rather than treating better health information principally as an underwriting threat, I argued that insurers should be thinking about how their products and services could evolve to help customers understand, manage and ultimately reduce their risks.
A decade later, that argument has become considerably more important.
The next stage may not simply be knowing that someone has an elevated probability of developing a disease. Increasingly, medicine may be able to intervene earlier, more precisely and eventually modify some of the underlying risk.
The Geneva Association is now highlighting exactly this broader issue. Its 2026 research identifies AI-supported diagnostics, genetic profiling, liquid biopsies, wearables, gene and cell therapies and other medical innovations as potentially changing morbidity, mortality, longevity and disability — the fundamental biometric risks underlying life and health insurance.[1]
But there is an even more intriguing development occurring in laboratories.
Can we change biological age?
In 2006, Shinya Yamanaka demonstrated that four transcription factors — OCT4, SOX2, KLF4 and c-MYC, now commonly known as the Yamanaka factors — could reprogram mature cells into induced pluripotent stem cells.
In simple terms, a specialised adult cell could have much of its cellular identity reset.
Researchers subsequently asked a fascinating question:
What if we don’t turn the clock all the way back?
Rather than completely reprogramming an adult cell into a stem-cell-like state, researchers are experimenting with partial reprogramming — transiently activating reprogramming factors with the aim of reversing some ageing-associated cellular changes while retaining the cell’s original identity.
The science remains experimental and substantial safety, delivery and cancer-risk questions remain. It would be wrong to suggest that we can currently rejuvenate people.
But laboratory and animal studies have already demonstrated changes in epigenetic age and cellular function. In one striking mouse experiment, expression of three reprogramming factors — OSK — promoted optic-nerve regeneration and restored aspects of visual function.[2][3]
Combine that field with AI-enabled drug discovery, gene and cell therapies, increasingly sophisticated prosthetics, neural interfaces, robotic surgery, 3D-manufactured implants and regenerative medicine.
The question for our industry becomes less fanciful:
What happens if ageing, disease and disability progressively become more measurable, treatable and, in some cases, repairable?
Picture a 65-year-old in 2040
They may look very different from today’s traditional retirement customer.
Their AI-enabled health system may have been monitoring biomarkers for years.
Cancer and cardiovascular risks may be identified much earlier.
Treatment may be increasingly personalised to their genome, biomarkers and disease characteristics.
A damaged joint may be repaired or replaced much more effectively.
Some previously disabling conditions may be mitigated through regenerative therapies, neural stimulation, advanced prosthetics or exoskeletons.
Their chronological age may be 65, while increasingly sophisticated measures of biological or functional age tell a rather different story.
Biological-age measurement is already attracting insurance attention. Munich Re, for example, is examining epigenetic and emerging multi-omic biological clocks, although current epigenetic clocks appear to add relatively little to established individual underwriting measures.[4]
By 2040, the distinction between chronological age, biological age and functional capacity could matter enormously.
And that creates some uncomfortable consequences for financial services.
The life insurance paradox
Better health should initially sound like very good news for life insurers. Earlier diagnosis and better treatment should reduce mortality. Regenerative medicine and better rehabilitation could reduce disability incidence or duration. Previously uninsurable conditions may become insurable.
But consider what happens when the customer knows their health has improved.
A healthy 60-year-old whose personalised medical information suggests a much longer life expectancy may decide they no longer require as much death cover.
They reduce or cancel it.
Someone whose health outlook has deteriorated has considerably more reason to retain theirs.
The remaining life insurance pool can therefore become progressively less healthy.
Now turn the proposition around.
The customer who believes they will live substantially longer has a very strong reason to purchase longevity protection.
They may cancel their mortality insurance while simultaneously retaining or increasing their lifetime income protection.
The insurer could therefore experience selection in opposite directions:
Better mortality risks leave the life pool.
Better longevity risks enter the annuity pool.
This isn’t simply better mortality experience. It potentially changes the economics of risk pooling itself.
The Geneva Association already identifies consumer access to sophisticated diagnostics as a potential source of information asymmetry and adverse selection.[1]
By 2040, that problem could be considerably more significant.
The product we need most could become the hardest to manufacture
If people live materially longer, longevity protection becomes more valuable. But it also becomes more expensive to guarantee. A lifetime annuity contains an extraordinarily simple promise:
We will continue paying you while you remain alive.
An insurer issuing such a promise to a 65-year-old today is potentially assuming medical-development risk over several decades.
What happens if mortality improves gradually? That can be modelled.
What happens if a combination of AI, cancer treatments, cardiovascular medicine and regenerative therapies produces a discontinuity? The difficulty isn’t simply predicting whether life expectancy increases by another two years. It is understanding the probability that it increases by five, ten or considerably more for particular cohorts.
That uncertainty ultimately has a price.
We may therefore reach the paradoxical position where consumers need longevity protection more than ever, precisely when unlimited lifetime guarantees become increasingly difficult and expensive for insurers to provide.
Interestingly, retirement systems are already exploring alternatives. A 2026 Society of Actuaries study identified 21 longevity risk-sharing programs internationally. Of the 20 analysed in detail, 19 principally shared longevity experience amongst participants rather than transferring the entire risk to an external insurer.[5]
Australia is simultaneously trying to develop its retirement-income market. APRA’s revised capital treatment for longevity products commenced on 1 July 2026, explicitly seeking greater innovation and competition in longevity products.[6]
The strategic question is whether the future lies predominantly in guaranteeing longevity risk — or pooling and sharing it differently.
What does “permanent disability” mean in 2040?
Disability insurance faces another challenge. Today we frequently treat damage, impairment and incapacity as closely related. Technology may increasingly separate them.
A person with spinal damage may remain medically impaired but regain significant mobility through neural stimulation and an intelligent exoskeleton. A lost limb may be replaced by an increasingly capable neurally controlled prosthesis. Damaged tissues may eventually become candidates for regenerative therapies. AI and robotic rehabilitation could improve recovery. This doesn’t make disability disappear. But it raises a fundamental question:
What exactly does Total and Permanent Disability mean when fewer disabilities are necessarily permanent?
Products may increasingly need to insure economic and functional incapacity, while insurers become much more actively involved in rehabilitation and restoration.
The implications extend well beyond life insurance into workers compensation, disability support and the NDIS.
Health insurance could face the opposite problem
Better medicine doesn’t necessarily mean cheaper medicine. In fact, initially it may mean precisely the opposite. AI may detect far more disease. And medicine may have something increasingly effective to do about it.
Gene therapies, personalised cancer treatments, cell therapies and regenerative medicine can create large upfront costs even where they generate substantial lifetime benefits.
The Geneva Association already identifies this tension: medical innovation can improve mortality and expand insurability while simultaneously creating significant pressure on health-insurance premiums and healthcare budgets.[1]
This creates another structural problem.
An insurer may pay $200,000 for an intervention today that saves hundreds of thousands of dollars in healthcare costs over the following twenty years. But the customer may move insurers next year. Our annual health-insurance model isn’t necessarily designed to finance interventions whose economic return emerges over decades.
And what happens to superannuation?
Perhaps the largest challenge isn’t insurance at all.
Our retirement system still broadly reflects a twentieth-century lifecycle:
What happens if healthy life is extended materially?
A healthy 70-year-old in 2040 may be entirely capable of working. But perhaps not five days a week. They may work for periods, retrain, consult, take extended breaks, care for family members, partially retire and return to employment. The boundary between accumulation and decumulation becomes increasingly artificial.
So too may a retirement system organised around particular chronological ages.
Financial advice also becomes harder.
Planning retirement to age 90 or 95 is one problem. Planning where someone’s plausible lifespan might range from 90 to 105 because we cannot know what medical innovations will arrive during the next thirty years is something else entirely.
Longevity becomes not simply a risk but a range of increasingly uncertain outcomes.
Asset allocation changes. Safe withdrawal assumptions change. The role of lifetime income changes. Inheritance occurs later. Housing decisions change. Growth assets may remain necessary much later in life.
We may eventually need to stop thinking about retirement planning and start thinking about whole-of-life capital and income planning.
Government has an even bigger problem
Australia already faces significant demographic pressure without any revolutionary medical breakthrough.
Treasury projects the proportion of Australians aged 65 and over reaching 23.4% by 2062–63 and estimates demographic ageing will account for around 40% of the projected increase in government payments, particularly through health, the Age Pension and aged care.[7]
But conventional long-term modelling necessarily makes assumptions about future mortality, health and technology.
What if those relationships change?
If people remain healthy for substantially longer, simply paying government pensions for additional decades becomes difficult to sustain.
The obvious response is increasing retirement ages.
Except another technological revolution is occurring simultaneously.
Medical AI could make humans capable of working longer.
Economic AI and robotics could reduce the amount of human labour required.
That creates a fascinating contradiction. We could have millions of healthy older people capable of productive work at precisely the point when economies need less conventional human labour.
Eventually that raises questions much bigger than superannuation:
How should economic output be distributed? Should taxation continue to rely so heavily on labour income? What happens to retirement ages? What happens to pension means testing? Who owns the productive capital created by AI?
And, in more disruptive scenarios, do concepts such as Universal Basic Income — or eventually some broader form of universal participation in AI-created prosperity — become part of mainstream economic policy?
These are longer-term questions, not predictions.
But executives designing systems intended to operate into the 2040s and 2050s should at least be thinking about them.
2040 isn’t very far away
This isn’t an argument that humans will shortly live to 150.
We don’t need to.
Even five additional years of lifespan combined with ten additional years of good health could materially change insurance, retirement and government economics.
Nor does every technology discussed here have to succeed.
Some undoubtedly won’t.
The important issue is the combined direction of travel. AI improves prediction. Medicine intervenes earlier. Genetics identifies risk. Cell and gene therapies modify disease. Regenerative medicine attempts to repair damage. Robotics and bionics restore capability. And epigenetic research raises the intriguing possibility that aspects of biological ageing itself may eventually become modifiable.
For more than a century, insurance has largely operated on the assumption that mortality, morbidity, disability and ageing are risks we observe, predict, pool and insure.
By 2040 they may increasingly become risks that individuals can measure, understand and sometimes modify. That requires more than better mortality tables. Boards, trustees and executives should be asking now:
- Are our products designed for this?
- Who should carry extreme longevity risk?
- How do we manage anti-selection when consumers may know more about their changing health than we do?
- Do today’s TPD and critical-illness definitions survive medical advances?
- How should health insurance finance expensive treatments with benefits extending over decades?
- Does a retirement system based around chronological age remain appropriate?
- And are our 30- and 40-year product strategies sufficiently flexible for medical developments we cannot yet predict?
The immediate challenge isn’t to predict precisely what medicine will achieve by 2040.
It is to ensure that products and systems being designed today can adapt if it achieves much more than our current assumptions expect.
An alternate ending…
There is, of course, a radically different view of 2040. Technologists including Elon Musk and Mo Gawdat envisage AI and robotics eventually creating such extraordinary productivity and abundance that the traditional relationship between work, income and retirement begins to disappear. Musk has argued that work could become optional, goods and services extraordinarily abundant, and governments could provide a “Universal High Income” rather than simply a Universal Basic Income; in his more extreme formulation, money itself could ultimately lose much of its relevance. Gawdat similarly argues that widespread automation will require new mechanisms such as UBI to give people purchasing power in an economy increasingly producing without human labour.
In that more utopian scenario, today’s debates about accumulating enough superannuation to fund retirement, purchasing longevity protection, or even preserving wealth for later life could eventually look very different. Whether that future is realistic is a discussion for another blog — but for an industry designing products that may still be paying benefits in 2040, 2050 and beyond, it is a scenario that should not simply be ignored.[8][9]
Endnotes
[1] The Geneva Association, Insuring Tomorrow’s Cures: Balancing the Promises and Practicalities of Innovative Medicine, February 2026. The report explicitly examines AI diagnostics, genomic profiling, gene and cell therapies and their potential effects on mortality, morbidity, disability, longevity, underwriting and adverse selection.
The Geneva Association report
[2] Cipriano et al., Mechanisms, pathways and strategies for rejuvenation through epigenetic reprogramming, Nature Aging, 2024. The review describes partial reprogramming research in mouse and human model systems and the significant unresolved challenges involved in translating it into treatments.
Nature Aging review
[3] Lu et al., Reprogramming to recover youthful epigenetic information and restore vision, Nature, 2020. The study demonstrated OSK-mediated effects on optic-nerve regeneration and visual function in mouse models.
Nature study
[4] Munich Re, Biological clocks: Ready for prime time?, July 2026. Munich Re notes growing interest in epigenetic biological-age measurement but concludes that current clocks add limited value beyond established underwriting factors, while multi-omic approaches may prove more useful.
Munich Re — Biological clocks
[5] Hanewald et al., Longevity Risk-Sharing Programs Around the World, Society of Actuaries Research Institute, June 2026. It identified 21 programs and found that 19 of the 20 analysed primarily use mutual participant risk-sharing rather than transferring longevity risk entirely to an insurer.
Society of Actuaries research
[6] APRA, Response to finalising amendments to the capital treatment for longevity products, March 2026. The new prudential standards commenced 1 July 2026 and are intended to support innovation and competition in Australia’s longevity-product market.
APRA longevity reforms
[7] Australian Treasury, 2023 Intergenerational Report. Treasury projects Australians aged 65+ reaching 23.4% of the population by 2062–63 and estimates ageing accounts for around 40% of projected growth in government payments.
Australian Treasury — 2023 Intergenerational Report
[8] Elon Musk: At VivaTech in 2024, Musk described a benign AI future involving “universal high income,” abundant goods and services and potentially optional work. He has subsequently gone further, arguing that continued advances in AI and robotics could eventually make money itself less relevant.
[9] Mo Gawdat: Gawdat has similarly argued that as AI increasingly performs productive work, existing economic and compensation systems will need to change, including potentially UBI, while also warning that abundance could coexist with extreme concentrations of wealth.
