I don't give a s*** about feeling cool—I'm here to make people feel safe.
As Americans face a retirement savings crisis deepened by potential Social Security cuts and rising costs, many are turning to AI chatbots for guidance on one of life's most consequential financial decisions. The technology offers a genuine entry point—running simulations, modeling scenarios, democratizing access to basic planning tools—but it carries within it the same blind spots and inherited biases of the financial culture it was trained on. What emerges is a portrait of a tool that can sketch the shape of a problem without yet being trusted to solve it.
- A single hypothetical question—can a 50-year-old woman with modest savings retire at 65?—yields three different answers from three different AI chatbots, exposing how inconsistent and assumption-laden the technology remains.
- With Social Security potentially facing a 20% benefit cut within six years and median retirement savings far below what experts recommend, the pressure on ordinary Americans to find reliable guidance has never been more acute.
- Experts warn that AI is trained on conventional Wall Street wisdom that may prioritize asset management over genuine financial safety, and that it routinely underestimates how long people will live and what long-term care will cost.
- The chatbots themselves, when pressed, admit to gaps they didn't surface upfront—no tax optimization, no long-term care modeling, planning horizons that stop at 90 rather than 100.
- Even if the technology improves, behavioral economists caution that fear of investing and financial anxiety—not lack of information—are what keep most Americans from engaging with retirement planning at all.
A 50-year-old woman with $185,000 saved and a $70,000 salary asks three AI chatbots whether she can retire at 65. Claude and ChatGPT say it's possible but precarious. Perplexity says probably not. When pressed, all three reveal they've been modeling her life only to age 90, ignoring long-term care costs and imprecise tax calculations. Claude, reconsidering its own math, walks back its initial optimism entirely.
This is the state of AI in retirement planning in 2026. About one in five Americans now turns to chatbots for financial advice, and among workers who already use AI professionally, half have applied it to retirement questions—twice the rate of those who don't. The appeal is understandable: retirement planning is brutally complex, most people are under-saved, and Social Security could cut benefits by 20 percent within six years if Congress doesn't act.
Some experts see real value in what AI can do. Certified financial planner Luke Delorme notes that chatbots can run Monte Carlo simulations—models that stress-test a portfolio across thousands of possible futures—which he calls 'the perfect thing for a computer program to do.' For someone without access to a human adviser, that's a meaningful starting point.
But Boston University economist Laurence Kotlikoff argues the technology may do more harm than good. AI is trained on conventional Wall Street wisdom, he says—advice built to keep assets under management, not to protect people from outliving their savings. It uses average life expectancy when it should use maximum life expectancy. MIT finance professor Andrew Lo adds that AI lacks tax optimization, regulatory nuance, and any legal obligation to act in a user's best interest.
Beneath the technical shortcomings lies a harder problem: many Americans are simply afraid of investing, parking savings in cash or CDs that lose ground to inflation. Delorme doubts a chatbot can overcome that anxiety. 'It's much more behavioral than it is a technical lack of knowledge,' he said. For now, AI can offer a sketch of an answer to retirement's hardest questions. Whether that sketch is accurate enough to build a life around remains very much unresolved.
A 50-year-old woman with $185,000 in retirement savings, earning $70,000 a year, asks three different AI chatbots the same question: Can she afford to retire at 65? The answers she gets are not quite the same. Claude and ChatGPT tell her it's possible but tight—she might run out of money. Perplexity is blunter: probably not, unless she cuts spending or works longer. When pressed on their assumptions, all three admit they're modeling her living only to 90, not 100. They're not calculating taxes precisely. They're not accounting for long-term care. And when Claude reconsiders its own math, it walks back its initial optimism entirely.
This is the state of artificial intelligence in retirement planning in 2026. About one in five Americans now uses chatbots for financial advice, and among workers who already use AI on the job, half have turned to it for retirement questions—twice the rate of those who don't use AI at work. The appeal is obvious. Retirement planning is the most consequential calculation most people will ever face, and it's also brutally complicated. The median worker with a retirement plan has saved far less than what experts say is needed. Social Security, the financial floor millions depend on, could cut benefits by as much as 20 percent within six years unless Congress acts. The pressure to find answers is real.
Some experts see genuine value in what AI can do. Luke Delorme, a certified financial planner in Massachusetts, points out that chatbots can run Monte Carlo simulations—mathematical models that test thousands of possible futures for a portfolio, accounting for bear markets and other shocks, then calculate the odds that your money will last. "It's the perfect thing for a computer program to do," he says. For someone with basic questions and no access to a human adviser, an AI tool might offer a useful starting point, a way to begin thinking through the numbers.
But the limitations are severe, and experts worry they're being understated. Laurence Kotlikoff, a Boston University economist who specializes in retirement, argues that AI may do more harm than good. The problem isn't just that large language models struggle with Social Security's 22,000 pages of rules or miss the tax implications of different withdrawal strategies. It's that they're trained on conventional Wall Street wisdom, which Kotlikoff sees as fundamentally misguided—advice designed to keep assets under management, not to keep people safe. When AI estimates how long you need your money to last, it typically uses average life expectancy. Kotlikoff says it should use maximum life expectancy, to guard against the catastrophe of outliving your savings. "I don't give a s*** about feeling cool," he said. "I'm here to make people feel safe."
Andrew Lo, a finance professor at MIT, adds another layer of concern. AI doesn't understand tax optimization. It doesn't grasp regulatory nuance. And unlike a human financial adviser, it faces no legal obligation to act in your best interest. When the CBS News team tested the three chatbots with the hypothetical 50-year-old woman, they discovered something telling: the apps themselves, when asked to explain their thinking, revealed gaps they hadn't mentioned in their initial advice. They weren't modeling the cost of long-term care, which can be enormous. They were using too short a planning horizon. The uncertainty was baked in, but it wasn't always surfaced upfront.
There's a deeper problem lurking beneath the technical limitations. Many Americans are afraid of investing. They keep their retirement savings in cash or CDs, which often earn less than inflation, slowly eroding their purchasing power. This isn't a knowledge problem—it's a behavioral one. Delorme thinks AI could help the roughly two-thirds of Americans who don't work with a financial planner begin to understand basic concepts. But he's skeptical that a chatbot can overcome the anxiety that keeps people from engaging with their finances in the first place. "It's much more behavioral than it is a technical lack of knowledge," he said. "I don't know if today that's going to help people overcome their fears."
The technology is improving. Experts expect Monte Carlo simulations and other computational tools to become more powerful. But the question of whether AI can truly serve as a retirement adviser—whether it can handle the full complexity of a person's financial life, with all its tax implications and Social Security scenarios and long-term care risks—remains open. For now, the chatbots can offer a sketch of an answer. Whether that sketch is accurate enough to live by is another matter entirely.
Notable Quotes
It's the perfect thing for a computer program to do. Eventually, I think that those tools will also become pretty powerful.— Luke Delorme, certified financial planner
It's much more behavioral than it is a technical lack of knowledge. I don't know if today that's going to help people overcome their fears of things, like the fear of investing.— Luke Delorme
The Hearth Conversation Another angle on the story
Why do you think people are turning to AI for something as important as retirement planning?
Because they're scared, and they're alone. Most Americans don't have a financial planner. They have a question that keeps them up at night—will my money last?—and a chatbot is free and available at midnight. It feels like an answer.
But the chatbots are giving different answers to the same question.
Right. And when you push them on why, they admit they're making assumptions they didn't tell you about. They're not modeling long-term care. They're not calculating taxes right. They're using a planning horizon that's too short. The technology is honest about its limits once you ask, but most people won't ask.
So the problem is that people trust them too much?
Partly. But also that the advice they're trained on—conventional Wall Street thinking—might be wrong to begin with. They're optimizing for the wrong thing. A human adviser might do the same, but at least they're legally required to act in your interest.
What would it take for AI to actually be useful here?
Transparency, for one. The chatbots need to lead with their limitations, not bury them. But also—and this is harder—they'd need to overcome the fact that most people's real problem isn't calculating numbers. It's being too afraid to invest, too anxious to engage. A chatbot can't fix that.
Can it get better?
The math will get better. The simulations will get more sophisticated. But the behavioral piece—the fear—that's human. That might always require a human to address.