Be Prepared: Retirement Could Be Drastically Impacted By AI
Expectations for artificial intelligence (AI) are running sky high—even for the near term. Focusing on the defined contribution landscape, Allspring sees several potential areas of impact over the medium and longer term.
Key takeaways
- Longer lifespans: As medical innovation increases life expectancy, retirement savings goals will also rise, along with the likelihood for additional regulatory changes.
- Increased litigation: AI may increase litigation related to retirement plans initially, but plan sponsors can also use AI to help avoid certain legal actions in the future.
- Investment personalization: Managed accounts may fill the data gap that’s burdened target date funds, which could help improve participant outcomes.
It remains to be seen whether AI will deliver on sky-high near-term expectations. In the medium to longer term, we see several potential effects of AI on the defined contribution (DC) landscape, including longer lifespans, increased litigation, and enhanced investment personalization.
Longer lifespans
As AI tools are applied to health care, previously undiscovered and newly personalized treatments could significantly boost life expectancy.1 Google DeepMind has already identified 2.2 million new materials, of which 380,000 may be able to be reproduced—that’s a tenfold increase above what scientists had previously identified.2
Paradoxically, medical successes brought about by AI could result in a failed retirement experience for many Americans. Imagine if our retirement period was lengthened by 7 to 10 years or even much longer. Science fiction may soon be reality. Whether that results in working longer, returning to work, living on less, or some combination thereof, the impact could be substantial. Our analysis suggests that adding just one year to an individual’s retirement period increases the risk of running out of money in retirement by 50%; adding three years leads to a dramatic 171% increase in the risk of running out of money.3 Longer lives could also delay the bequest of assets to the next generation, which historically has helped millions of Americans fund their non-working years.4
Retirees’ ability to meet their needs is a critical component of a well-functioning society. Fortunately, major shifts in the U.S. retirement system are not new. Life expectancy has increased in the past, and sources of retirement income have shifted. Our modern idea of retirement became popularized in the 1920s and 1930s as living standards rose and Social Security was introduced. Other major changes along the way included the now 50-year-old passage of ERISA, which led to the rise of the 401(k), and the 2006 Pension Protection Act, which enabled automatic and default features. Recently, we’ve seen Congress pass SECURE 1.0 and SECURE 2.0, as defined benefit plan coverage has declined and DC plans have taken center stage.
If AI leads to gradual increases in longevity, we may well see Congress pass SECURE 3.0, 4.0, 5.0, and beyond. If life expectancies increase dramatically over a short period, however, this could require a fundamental rethink of the concept of retirement more akin to the 1935 establishment of the Social Security system.
Bottom line: AI could gradually or radically increase lifespans, making DC plans even more crucial and potentially inviting government action or even a rethink of the broader concept of retirement.
A spike in litigation … and better ways to avoid it
AI adds an incredible capacity to digest documents and text. For example, Allspring has been using AI for years to sift through tens of thousands of news articles every day to gauge sentiment ahead of companies’ dividend announcements.5 Unfortunately, AI could also be used in ways that plan sponsors may not appreciate. That includes plaintiff lawyers identifying plans at risk for litigation by scraping data from websites and from regulatory filings such as Form 5500. AI may also be used to help identify disaffected participants (current or terminated) who might be open to filing a suit. Litigation could initially spike, especially for medium-to-small plans that aren’t as highly scrutinized today.
It’s not all bad news. While plaintiff lawyers may be the first movers, plan sponsors should be able to hit back. AI can suggest best practices to avoid litigation, and it may save time by filling out regulatory documents quickly and in ways that don’t attract lawsuits. Many plan sponsors will benefit from enhanced advice from their consultants and advisors; some may also consider pooled employer plans (PEPs). Plan sponsors with dedicated internal DC resources could test new ideas more quickly and disseminate detailed implementation steps to their peers.
Bottom line: There may be an initial rise in lawsuits, but AI may also help plan sponsors fight back. Many plan sponsors will likely find the quality and value of advice to be even more important than it is today, with AI driving innovation and best practices.
Investment personalization
Personalization has been a key trend for more than a decade, even if it hasn’t made many headlines. Most of us are so used to getting recommendations for new shows and music that we hardly notice it. We take for granted that online retailers promote products based on past purchases.
By contrast, target date funds were invented in the mid-1990s. They do an admirable job of optimizing the stock/bond risk allocation over a participant's life, but they rely exclusively on one known data point—expected retirement age—and make many additional assumptions that may not be accurate for a given individual. Managed accounts, meanwhile, rely on more data and may provide more appropriate asset allocation depending on a participant’s unique situation. Managed accounts are growing quickly at 13% annually; consultants expect faster growth than any other investment option over the next 12 months.6
As AI becomes more integrated into DC plans, the growth of managed accounts may significantly accelerate—not only as an option on the menu but also increasingly as the default. Two reasons include:
- Better inputs: As AI chatbots become integrated into DC plans, they offer an opportunity to have a personalized “conversation” with participants, asking deeper questions that might not be captured in a standard questionnaire. Examples could be getting a better picture of a participant’s total wealth and tax situation.
- Better outputs: With access to data from millions of participants, AI will be able to build better personas, making assumptions about the financial situation of individual participants based on data on their peers and helping construct portfolios that are specific and personalized—for example, diversifying away from the industry in which the participant works or putting tax-inefficient investments in the plan to offset more tax-efficient out-of-plan investments.
Personalization may go far beyond what DC experts are thinking about today, incorporating a total asset/liability picture, career trajectory, family, health data, and more to provide a highly customized experience—like a personalized CFO.
Bottom line: With so much data, and AI’s ability to process it, greater personalization seems inevitable. In particular, AI may turbocharge managed accounts, supplanting target date funds over time and improving participant outcomes.
Final takeaways
It’s still too early to know how AI will affect DC plans and society as a whole, but one thing is clear: Allspring is not alone in expecting big medium- to long-term changes. A recent example is Microsoft’s proposed 20-year agreement to reopen the Three Mile Island nuclear power plant, paying $1.8 billion to use all the power to feed its yet-to-be-built AI data centers.7 Google is pursuing a similar strategy, signing a contract to purchase nuclear energy produced by multiple smaller reactors to fuel its future AI power needs.8
Allspring believes it’s critical to remain intellectually humble about AI. We expect events will unfold in ways that have yet to be envisioned. Still, plan sponsors would do well to gain a deeper understanding of AI and managed accounts, fortify themselves against litigation, and consider a future in which participants live much longer than they do now.
For more on DC trends and research, reach out to your contact at Allspring.
1. The Economist, “Ray Kurzweil on how AI will transform the physical world,” June 17, 2024, accessed at https://www.economist.com/by-invitation/2024/06/17/ray-kurzweil-on-how-ai-will-transform-the-physical-world
2. Wired Magazine, “Google DeepMind’s AI Dreamed Up 380,000 New Materials. The Next Challenge Is Making Them,” November 29, 2023, accessed at https://www.wired.com/story/an-ai-dreamed-up-380000-new-materials-the-next-challenge-is-making-them/#:~:text=Using%20that%20information%2C%20the%20AI,stable%20materials%20nearly%2010%2Dfold.
3. Allspring. For more on this, ask your Allspring contact about our research, “Retirement Readiness: What Is Success?”
4. Jeffrey R. Brown, Courtney C. Coile, and Scott J. Weisbenner, “The Effect of Inheritance Receipt on Retirement,” The Review of Economics and Statistics, 92 no. 2 (May 2010) pp. 425–434.
5 Sophie Careford, Eddie Cheng, and Vince Fioramonti, “Enhance Outcomes With AI and Machine Learning,” February 21, 2024, Allspring, accessed at Enhance Outcomes with AI and Machine Learning | Allspring Global Investments.
6. Cerulli Edge, U.S. Retirement Edition, 3Q 2024.
7. Reuters, “Microsoft deal propels Three Mile Island restart, with key permits still needed,” September 21, 2024, accessed at https://www.reuters.com/markets/deals/constellation-inks-power-supply-deal-with-microsoft-2024-09-20/.
8. The Associated Press, “Amazon, Google make dueling nuclear investments to power data centers with clean energy,” October 16, 2024, accessed at: https://apnews.com/article/climate-data-centers-amazon-google-nuclear-energy-e404d52241f965e056a7c53e88abc91a.