Reflections on AI for Aging and the Social Side of Sustainable Investing

At CES, one of the largest events in the tech industry, I noticed that many older attendees seemed confused when trying to use AI longevity products. I noticed that AI longevity products were almost never built with data from older adults. Training sets came from younger users, design teams skewed young, and seniors appeared mostly as the imagined end customer rather than as a real source of input. How can capital claim to serve the longevity economy when the people that the economy targets are barely visible in the data?

Our client works on AI for aging in East Asia, where demographic pressure is high and product cycles are fast. Our team is helping design a platform that connects product teams with verified older users so that real first-party data can feed back into the design process. Working through an iceberg exercise in this market, we kept moving from surface events like product launches and policy announcements down to the mental model underneath, and the pattern at the bottom, which shows that older people are positioned as peripheral rather than central actors. Since the dataset is not from the actual elderly, these products are hard to use in the actual daily life. The private capital hesitated to invest due to unclear profitability, then only the government puts effort and money to these projects. Now the model is still working because of government support. Once the government leaves, everything changes, the whole AI longevity business circle gets stuck here. Thus, we’re trying to figure out a sustainable investing problem rather than only a product design problem.

Two papers helped me see why this matters as a sustainable investing question rather than only a product design one. Amel-Zadeh and Serafeim (2018) surveyed mainstream investment professionals and found that the most common reason investors use ESG information is relevance to investment performance, and that the social dimension of ESG is the hardest to operationalize, mainly because reporting is inconsistent and comparable data is scarce. Their result reframed my project in a useful way. The core problem is not that investors refuse to care about how older adults are served. It is that the data they would need to have to care, is not there to begin with. Research on digital ageism in AI (Chu et al., 2022) documents the same scarcity at the product level, with older adults underrepresented in datasets and design teams.

If Amel-Zadeh and Serafeim name the data problem, Boulongne, Durand, and Flammer (2024) show what tends to happen when that problem is fixed. Writing in the Strategic Management Journal, they study impact loans made to business ventures inside and outside French banlieues, neighborhoods that traditional lenders systematically underserve. Loans directed into banlieues produced larger gains in both financial performance and social impact, including local employment, job quality, and gender-equitable hiring. A controlled experiment with working professionals showed that loan officers rejected identical ventures more often when they were in banlieues, which helps explain why the market underprices these opportunities. Their argument is not that impact investing is charity done better, but that it identifies pockets of unrealized potential that traditional capital misses because of information and bias problems.

The two papers together changed how I see our platform. The AI-for-aging market has its own version of the banlieue problem. Private capital stays on the edges not because the underlying need is absent, but because the information needed to price risk and demand responsibly is missing. In that light, the platform is less a standalone B2B product and more a piece of market infrastructure. It is the kind of infrastructure that would let impact investors, blended finance structures, and later-stage private capital enter a market that is currently almost entirely public-payer dependent, and that would let the resulting products earn the claim of social impact rather than assume it.

The hardest work in sustainable investing for the longevity economy is upstream of the capital itself. It sits in who is counted, how they are counted, and whether the resulting information is credible enough for markets to act on. If that upstream work is done well, the capital structure downstream has a chance to stop depending on the next public budget cycle. If it is not, no amount of financial engineering will make the loop close.

References

Amel-Zadeh, A., & Serafeim, G. (2018). Why and how investors use ESG information: Evidence from a global survey. Financial Analysts Journal, 74(3), 87–103. https://doi.org/10.2469/faj.v74.n3.2

Boulongne, R., Durand, R., & Flammer, C. (2024). Impact investing in disadvantaged urban areas. Strategic Management Journal, 45(2), 238–271. https://doi.org/10.1002/smj.3544

Chu, C. H., Nyrup, R., Leslie, K., Shi, J., Bianchi, A., Lyn, A., McNicholl, M., Khan, S., Rahimi, S., & Grenier, A. (2022). Digital ageism: Challenges and opportunities in artificial intelligence for older adults. The Gerontologist, 62(7), 947–955. https://doi.org/10.1093/geront/gnab167