Chasing brighter futures: AI and retirement plans 

Lifespans and the cost of living are rising beyond what retirement income systems were built to support. Covering basic expenses could require people to work longer, save and invest more aggressively, or adjust their standards of living. 

How might we ensure comfort and dignity in retirement? In its Longevity Economy Principles report, the World Economic Forum in collaboration with Mercer reimagines the role of jobs, health and other drivers in building long-term financial resilience¹. To that end, finance and benefit leaders are exploring AI’s potential to blunt the looming crisis by improving modern pension systems, developing smarter investment practices, and designing better employee benefits programmes. In fact, they’ve already begun. 

Investment managers have used AI for years, analysing data to forecast predefined trends and behaviours for better decision-making. More recently, generative AI (gen AI) has emerged with a host of new capabilities that could also potentially enhance retirement plans and benefits: 

• Identifying subtle connections and inferences from big data 

• Integrating different datasets for more robust, high-quality insights 

• Creating novel multimedia content at high speed and scale 

• Serving as autonomous agents to handle tasks and interactions  

Solving the retirement crisis will take more informed strategies and decisions, yet there are countless variables to consider. Each retirement income system evolves from distinct social, historical, economic and political factors. And while these drivers are often interconnected, the ties that bind them aren’t always linear — or even apparent. 

Through a mix of predictive and generative AI, we may be able to better understand the drivers, features and potential outcomes of different retirement plans. One helpful starting point is the Mercer CFA Institute Global Pension Index, which (without the use or consideration of AI) rates these income systems on a range of variables in three key areas: adequacy, sustainability and integrity. 

Adequacy is the extent to which current retirement income provides financial security for today’s retirees. In many countries, this largely depends on private pension plans — both how members engage with them, and how they’re designed. These plans have six design features that can improve retirement income adequacy, and AI could soon help members use them effectively. Some opportunities for AI to improve retirement adequacy include: 

• Taxation support. AI could optimise the value and impact of tax incentives for plan participation. 

• Preservation. AI could curb benefits leakage by suggesting a minimum access age. 

• Vesting/portability. AI could analyse historical data to model different vesting and portability scenarios. 

• Retirement benefit design. AI could propose the optimal mix of lump-sum and annuity payments. 

• Separation. AI could suggest a fair and equitable division of pension benefits during a divorce. 

• Continued accrual. AI could assess the impact of benefits accruing when workers can’t work for a variety of reasons. 

It remains to be seen whether changing demographics and economics will affect the long-term viability of pensions and social security systems. Some people could outlive their retirement savings by nearly 20 years unless they remain in the workforce for longer, or receive more support from these income systems². Further, while the rising old-age dependency ratio could mean we’ll have a shrinking workforce supporting more retirees, it’s also based on a retirement age of 65 that seems increasingly outdated as more people work into their late 60s or 70s³. Add in other factors such as inflation and government debt, and it’s at least clear that the future holds a great deal of uncertainty. 

Artificial intelligence has the potential to help facilitate more sustainable pensions and social security programmes. It may be able to improve both productivity and labour market access across different age groups, unlocking more ways for people to remain engaged in the workforce for longer. AI also has the potential to help governments and administrators analyse past and present data around consumer expenses, member behaviour, contribution rates and economic growth. The learnings could then inform more future-focused decisions, such as raising the state pension age or facilitating phased retirement opportunities, to possibly improve these systems’ financial sustainability.

Effective retirement income programmes need public participation, which in turn depends on the public’s trust in these systems. Running these plans with integrity means ensuring that members gain sufficient value at a reasonable cost while keeping them informed. And given the increase in defined contribution (DC) plans, that transparency should extend to each plan’s investment strategies and performance. 

AI could have the potential to boost plan integrity by driving strategies that aim to control costs, improve outcomes and optimise member communications. It could streamline and automate transactional tasks to make plan administration more affordable. Gen AI could personalise member interactions for better value and understanding while standardising technical and compliance-driven language for greater consistency. 

Ethics and integrity go hand in hand, and the potential use of AI in pensions and social security poses a number of ethical challenges. Could AI unfairly restrict retirement solutions for certain members or groups? Will the public have a say in the use and development of AI tools that could potentially impact their financial well-being? And in the event of unhelpful or misleading financial advice from AI, who would be held responsible?

Governance and human oversight can preserve the public’s trust in retirement income systems, especially in the age of AI. Experienced pension professionals can help ensure equity, transparency and accountability in how AI-powered retirement solutions would be used and developed. Their input could help validate AI models’ predictions and recommendations, as well as potentially mitigate the risks of adverse decisions or outcomes.

Investment returns are an essential source of retirement income. These returns depend on several factors such as investment strategies, economic conditions and government regulations. 

Many investment professionals are exploring AI’s future potential. Mercer research shows that 91% of investment managers are either using (54%) or planning to use (37%) AI for investment strategies or asset class research. Further, more than two in five CFA Institute members already use AI and big data to support task automation, smarter decisions and higher-quality offerings. 

So, can AI boost investment returns for better retirement outcomes? Consider the following use cases: 

• Analysing data, risks and market trends to inform investment decisions 

• Revealing hidden patterns and market signals for better asset allocation and diversification 

• Powering investment products that better align with investors’ values 

• Monitoring and addressing various investment-related risks in real time 

• Predicting member behaviours that may impact cash flows, such as asset reallocation 

To be sure, the evolution of AI poses new challenges in the investment space. Low-quality data, hallucinations and bad actors can all cause AI to fuel misinformation and disinformation, the number-one risk in the World Economic Forum’s Global Risks Report⁴. Cyber risk is another key concern, given the sensitive data in this space and the potential for damage if that data becomes compromised. Because these forces could drive flawed decisions and even financial loss, it’s important to stay vigilant through robust governance plans, enhanced data security and future-focused regulations as AI gains traction. 

People face complex financial decisions around retirement. It starts while they’re still in the workforce: which plans or programmes to choose, how much to save or contribute, and ultimately, the retirement goals they’re working toward. Behind these decisions are several factors to consider, such as age and career stage, assets and income sources, dependants and partners, and health and well-being. 

Most workers don’t expect to make these important long-term decisions alone. Today, just 45% of employees globally believe they are saving enough for the future⁵ — and two-thirds are trusting their employers to help them plan for retirement. To be sure, these numbers vary by country: Mercer’s Global Talent Trends 2024 reveals that workers in Australia (71%), India (71%) and the US (72%) are much more likely than peers in Hong Kong (62%), Japan (49%) and Singapore (50%) to expect their employers’ support. But across the globe, fewer than two in five executives (38%) see retirement benefits as a growth driver, and just 18% think it’s important to understand different workers’ retirement adequacy. 

Employers can use AI to provide more data-driven education for employees, as well as cost-effective benefits programmes that have the potential to improve overall financial well-being. AI could help analyse each individual’s situation, guide them all through the benefits ecosystem, and craft benefits communications that keep employees informed throughout their tenure. According to Global Talent Trends 2024, a majority of HR leaders are either using or planning to use AI in related processes:

• Predicting when older workers with critical skills are likely to retire (77%) 

• Making recommendations to help employees select the right benefits (78%) 

• Helping workers advance their financial well-being (77%) 

Trusted financial advice can be cost-prohibitive, and underserved populations — such as low-wage or self-employed workers — may face the added challenge of making these decisions alone. In the future, AI-powered tools and insights could potentially improve information access and retirement outcomes for these groups. Governments and financial advisers will continue to have an important role to play in ensuring these segments of the workforce don’t get left behind in the age of AI. 

AI could potentially support modern retirement income systems, future-fit investment decisions and stronger employee benefits for lasting financial wellness. Organisations are exploring its capacity to unlock new opportunities across the retirement industry, such as more efficient client servicing, personalised member engagement in real time, and more informed decision-making in plan design and administration. 

Yet even in the face of a looming retirement crisis, there’s significant risk in trusting AI with too much too soon. All the systems and efforts that shape our health, wealth and careers demand human expertise across sectors and functions, increased data security to neutralise cyber risk and robust governance plans to better comply with evolving regulations. 

AI is advancing every day, however. And while there’s still a long way to go, the finance and benefits leaders who explore it today will be firmly positioned tomorrow to reap the rewards. For more on AI from an investments perspective, download our recent report — and for a closer look at retirement plans beyond the black box of AI, explore the Mercer CFA Institute Global Pension Index.

Footnotes

(1) World Economic Forum. Longevity Economy Principles: The Foundation for a Financially Resilient Future, 2024. Available at https://www.weforum.org/publications/longevity-economy-principles-the-foundation-for-a-financially-resilient-future/
(2) Nazeri H. “People are living longer. So how can we build resilient economies and thriving societies?” Available at https://www.weforum.org/agenda/2024/01/longevity-economy-thriving-societies/
(3) OECD. Old-age dependency ratio, 2024. Available at https://www.oecd.org/en/data/indicators/old-age-dependency-ratio.html
(4) World Economic Forum. Global Risks Report 2024. Available at https://www.weforum.org/publications/global-risks-report-2024/
(5) Whiting K. “We’re living longer. Here’s how that will change retirement,” 2023. Available at https://www.weforum.org/publications/global-risks-report-2024/