Two out of three American adults have now asked a chatbot for financial advice. Let that sink in for a second. The same technology that can’t always tell you who won the 1994 World Series is now fielding questions about your retirement savings.

A new Credit Karma survey puts the number at 66% of adults using AI for financial guidance. Among millennials and Gen Z, it jumps to 82%. And while those younger generations might have decades to recover from a bad tip, the trend isn’t staying in their lane. More and more people nearing or in retirement are opening up ChatGPT and typing in questions that used to go to a financial advisor — or at least a trusted friend who reads the Wall Street Journal.

What AI Actually Does Well — And Where It Falls Apart

Here’s the thing: AI can be genuinely useful for the basics. Need to understand what a required minimum distribution is? Want a quick breakdown of how a Roth conversion works? It’ll give you a decent answer in about four seconds. That’s faster than Googling, and usually clearer than the IRS website — which, let’s be honest, isn’t a high bar.

CBS News business analyst Jill Schlesinger put it well:

“I think AI can be great for general education. How do I read my pay stub? What’s the difference between a Roth and a traditional IRA? But when you’re making a big decision it is kind of scary to rely on AI which can hallucinate, and I have to also say that there is a concern about who much you’re sharing on AI for privacy.”

She’s not wrong. AI doesn’t know your tax situation, your health, your spouse’s pension, or whether your roof needs replacing next year. It doesn’t know that you’re carrying $14,000 in credit card debt or that your daughter might need help with a down payment. It gives you general answers to specific problems — and in retirement planning, the specifics are everything.

Here’s What That Actually Means For You

If you’re retired or within striking distance, the stakes of a wrong answer are higher than they are for a 28-year-old with decades of paychecks ahead. Pull money from the wrong account in the wrong order, and you could trigger a tax bill that eats into savings you can’t replace. Misjudge your Social Security timing by a year or two, and you’re leaving real money on the table — potentially tens of thousands of dollars over your lifetime.

AI won’t warn you about any of that unless you know exactly the right question to ask. And if you already knew the right question, you probably wouldn’t need the help.

There’s also the privacy piece. Typing your income, account balances, and Social Security number into a chatbot is handing personal data to a system you don’t control. Experts advise reviewing privacy policies, adjusting your settings, and never sharing information you wouldn’t want stored on a server somewhere.

How To Use This Without Getting Burned

None of this means you should avoid the technology entirely. Used right, it’s a solid research assistant. Here’s how to keep it in its lane:

Use AI for learning, not for decisions. Ask it to explain concepts, compare account types, or walk you through how tax brackets work. Don’t ask it whether you should roll over your 401(k) — that answer depends on about fifteen things it doesn’t know about your life.

Keep your personal details out of the conversation. You can ask smart questions without typing in your account numbers or Social Security info. Treat it like a library book, not a confessional.

Verify before you act. AI can sound confident and still be dead wrong. If a chatbot gives you a number — a tax rate, a contribution limit, a benefit amount — check it against an official source before you make a move.

The Bottom Line

Technology reshaping how Americans handle money isn’t new. We went from passbooks to online banking to mobile deposits. AI is the next step, and it’s moving fast. The question isn’t whether you’ll use it — odds are, you already have. The question is whether you’ll use it as a starting point or a finish line. For anyone protecting a nest egg they spent forty years building, that distinction matters more than any algorithm.