Two surveys released this year found the same odd split: Americans have adopted AI broadly, and then stopped at the door of their own finances.
The National Institute on Retirement Security's Retirement Insecurity 2026 report, released August 26, found 63% of respondents had used an AI tool — but 61% had never used one for a financial, investment, or retirement question. 45% said they were uncomfortable with AI playing any role in delivering financial advice. Interest was highest for the lowest-stakes task: 38% for budgeting, 34% for investing, 32% for retirement planning.
A Gallup poll of 5,075 adults, fielded March 20 to April 6, points the same direction. About one in five people who sought financial advice in the past year turned to AI. Only about three in ten expressed a great deal or some confidence in AI's financial expertise, and just 3% said they trust it a great deal.
The generational gradient is steep, and it runs against the people with the most at stake. Roughly 25% of Gen Z and millennials used AI for financial advice, versus 16% of Gen X and 7% of baby boomers — the cohort closest to drawing down what it has saved.
The Evidence Complicates the Skepticism
A 2026 study in the Financial Planning Review by Efthymia Antonoudi and Eric Ludwig of The American College of Financial Services examined 2,000 state and local government employees and found that every advice channel was associated with higher odds of saving for retirement than no formal guidance at all. Consulting a financial planner showed the strongest single effect — reported as 181% higher odds of saving. AI-only users also saved at higher rates than the unadvised. The strongest association of all was for people using both.
That is the finding worth sitting with. The comparison that matters for most households is not AI versus a planner. It is AI versus nothing — which is what a large share of savers currently have. NIRS found 47% have less than $100,000 saved and 18% have nothing at all.
Where It Breaks
The Center for Retirement Research at Boston College is direct about the limits: AI "can still provide misinformation when concepts are too nuanced or when inputs are incomplete," and it performs best on well-defined questions with clear frameworks rather than tangled personal scenarios. MIT Sloan's Taha Choukhmane offered a narrow endorsement in Gallup's coverage — he would encourage people "to use AI to explain and define" financial concepts.
Explain and define. Not decide.
Practical Takeaways
- Use it as a translator, not an advisor. "What is the pro-rata rule?" is a good question. "Should I do a backdoor Roth?" depends on facts the tool does not have.
- Verify every number against the primary source. Contribution limits, RMD ages, and IRMAA brackets change annually, and models confidently repeat stale figures. Check IRS publications and ssa.gov directly.
- Treat incomplete inputs as the main failure mode. AI answers the question you asked. It will not tell you that your real problem is a pension survivor election or a state tax rule you never mentioned.
- Do not paste account numbers, Social Security numbers, or statements into general-purpose tools. Data privacy on linked financial accounts is an unresolved concern flagged by CRR's readers.
- Use it to prepare for a human, not replace one. The strongest result in the research was the combination. Drafting your questions with AI before meeting an advisor is a legitimate way to buy better use of billable time.
- Skepticism is not a plan. If the alternative to imperfect help is no help, the savings data does not favor waiting.
Sources: National Institute on Retirement Security, Retirement Insecurity 2026 (August 26, 2026); Gallup poll on AI and financial advice (March 20–April 6, 2026); Antonoudi & Ludwig, Financial Planning Review (2026); Center for Retirement Research at Boston College.