AI Can Answer Almost Every Question About Your Money. Except the One That Matters Most.

The problem isn't your prompt. It's the data that doesn't exist anywhere AI can reach.

There is a moment happening right now in millions of households across America.

Someone opens ChatGPT or Claude or Gemini. They type a question about their finances. They get back a confident, well-structured, authoritative-sounding answer. And they feel — for a moment — like they finally got the guidance they needed.

That moment is real. And for certain financial questions, AI genuinely delivers.

But there is one question — the question most retail investors actually need answered — that no AI tool can touch. Not because of a bad prompt. Not because of hallucination. Not because the model isn't smart enough.

Because the data required to answer it has never been made available to anyone outside of Wall Street.

What AI is actually good at

In a recent CNBC article, personal finance reporter Greg Iacurci spoke with Andrew Lo — Director of MIT's Laboratory for Financial Engineering and principal investigator at MIT's Computer Science and Artificial Intelligence Lab — about the growing use of AI for personal finance guidance.

Lo's take was balanced and credible. AI is genuinely useful for high-level financial education. Why diversification matters. How ETFs compare to mutual funds in different scenarios. The mechanics of compound interest. General retirement planning frameworks.

The numbers back this up. According to a recent Intuit Credit Karma survey, 66% of Americans who have used generative AI say they have used it for financial advice. Among millennials and Gen Z that number exceeds 80%. About 85% of those who received AI financial guidance acted on it.

For financial education — for understanding concepts, frameworks, and general principles — AI is a useful and improving tool. Lo, who has studied this extensively, says people should be using AI for financial planning. The question, he told Iacurci, is how.

Where AI hits a wall

Lo is equally direct about AI's limitations.

"When it comes to very, very specific calculations of your own personal situation, that's where you have to be very, very careful," he told CNBC.

AI struggles with precise numerical analysis of individual situations. It can hallucinate — producing answers that sound authoritative but are simply wrong. And as Lo noted: "One of the things about large language models that I find particularly concerning is that no matter what you ask it, it'll always come back with an answer that sounds authoritative, even if it's not."

Brenton Harrison, a certified financial planner quoted in the same piece, put it simply: "Even if it's the best model in the world, if it's fed a bad prompt it will only be able to do so much."

No matter how carefully you craft your prompt, AI is working only with what you give it. Which leads directly to the question it cannot answer.

The question no AI can answer

Here it is:

Is my portfolio actually performing well — compared to investors exactly like me?

Not compared to the S&P 500. Not compared to a theoretical model portfolio. Not compared to some generic benchmark constructed by a committee.

Compared to real investors. Investors who hold the same mix of stocks, bonds, ETFs, and mutual funds you hold. Investors in the same risk category. At the same type of firm. What did they actually take home?

That is the question that has driven retail investor anxiety for thirty years. It is the question behind every panicked search, every midnight Reddit post, every uncomfortable conversation with a financial advisor that ends in vague reassurance rather than a clear answer.

And AI cannot answer it. Not because of prompt engineering. Not because the model isn't sophisticated enough. But because the data required to answer it — verified, anonymous, cross-custodial, real-world performance data from investors like you — has never been made available outside of institutional finance.

Why the S&P 500 is the wrong benchmark for most retail investors

Most retail investors have been so thoroughly conditioned by the industry to use the S&P 500 as their benchmark that they've never questioned whether it actually applies to them.

It doesn't. For most of them, it doesn't.

The S&P 500 is a scoreboard for a very specific game — the performance of 500 large-cap U.S. companies selected by committee. If your portfolio holds bonds, international stocks, small-cap or mid-cap positions, or any mix other than pure large-cap U.S. equity — and the overwhelming majority of retail investors do — then comparing your performance to the S&P 500 is not just unhelpful. It is actively misleading.

You are measuring yourself against a standard that was never designed to measure what you own.

DALBAR has documented this mismatch for decades. Their annual Quantitative Analysis of Investor Behavior consistently shows that retail investors feel like they are failing against a benchmark that was never designed to measure their actual portfolios. In 2024 alone, the average equity fund investor trailed the S&P 500 by 848 basis points — despite the market delivering a 25% return. Investors guessed the market's direction correctly only 25% of the time, tying a record low.

The problem is not lack of information. Retail investors have more financial information available today than at any point in history. The problem is the absence of a specific kind of data — peer data. Objective, verified performance data from investors who are genuinely comparable to them.

No AI prompt, however well crafted, can fill that gap.

What a true peer benchmark actually is — and why it's the most powerful tool in personal finance

Imagine you could aggregate every 80% stock / 20% bond portfolio that exists. Every one. Across every brokerage, every firm, every custodian. And you could calculate what that entire universe of investors actually earned — not what a model says they should have earned, not what an index returned, but what real investors with real portfolios took home after all the friction of real life was paid.

That would be the true benchmark for an 80/20 investor. Not because it's the largest possible dataset — though size matters — but because it is the only comparison that is genuinely apples to apples. Same risk profile. Same real-world constraints. Same market conditions. The only variables are which firm is managing the money and which specific holdings were chosen.

That comparison answers the question that no index can answer: given exactly what I own and exactly how much risk I'm taking — how am I actually doing?

And here is what most people don't realize: you don't need the entire world to make it real.

The statistical science on this is settled. Using established confidence interval methodology, 250 portfolios within a defined risk category delivers 93.8% benchmark accuracy at a 95% confidence level. That is the standard used in peer-reviewed research and institutional benchmarking. It is the floor at which a benchmark becomes statistically defensible — not a guess, not an estimate, but a verified measurement.

Two hundred and fifty people. That is all it takes to give every investor in that category something the industry has never given them. A real answer to the real question.

When you know where you actually stand — not against an index you don't own, but against investors who hold exactly what you hold — the noise stops being noise. A market downturn is not a personal failure. It is a shared condition. The question is not whether your portfolio went down. The question is whether it went down more or less than your true peers. That is the only comparison that tells you anything real about whether your money is being managed well.

The investor who discovers their firm ranks in the bottom third for clients in their risk category has actionable information. The investor benchmarked only against the S&P 500 has nothing but a feeling.

The difference between information and data

This is the distinction that matters most.

AI has access to enormous amounts of financial information. Principles. Frameworks. Historical market data. Academic research. General guidance. It can synthesize that information faster and more clearly than any human advisor.

But information is not data.

The specific, verified, real-world performance of investors in your exact risk class at your exact firm — that is data. And it doesn't exist in any AI training dataset. It doesn't exist on the public internet. It has never been aggregated, anonymized, and made available in a way that any AI tool — or any retail investor — could access.

Wall Street has this data. Institutional investors have this data. The firms managing your money have this data. They use it every day to evaluate performance, benchmark results, and make decisions.

Retail investors — the people who actually generated the data — have never had access to it.

That is not an oversight. That is a structural feature of an industry that benefits from keeping performance opaque.

What changes when the data exists

Pure Benchmarks was built to change that. Not to tell you what to do with your money. Not to give you a model portfolio or a projection or a theory. Just to give you what you were always owed — the objective truth about how your money is actually performing relative to the people who are most like you.

Pure Benchmarks does have an AI feature — Hetty — coming soon. But Hetty is different from every other AI financial tool in one fundamental way. She doesn't need better prompts to answer whether your portfolio is performing well relative to your peers. She has the peer data to answer it. Verified. Anonymous. Real.

That is not a prompt engineering advantage. That is a data advantage. And it is the only kind of advantage that actually answers the question that has mattered most for thirty years.

The right tool for the right question

AI is genuinely useful for financial education. Use it. Ask it to explain concepts. Ask it to walk you through the mechanics of dollar-cost averaging. Ask it why the S&P 500 is the wrong benchmark for most retail investors. For those questions it performs well and gets better every day.

But when you need to know whether your money is actually being managed well — when you need a real answer to the real question — you need data that AI simply does not have.

That data now exists. And for the first time it belongs to the investors who generated it.

Reporting by Greg Iacurci, CNBC Personal Finance Reporter, April 18, 2026: "There's an 'art' to writing AI prompts for personal finance, MIT professor says." Expert quoted: Andrew Lo, Director of MIT's Laboratory for Financial Engineering. DALBAR data referenced from the 2025 Quantitative Analysis of Investor Behavior report. Survey data from Intuit Credit Karma, September 2024. Statistical methodology based on Cochran's formula at 95% confidence interval. This post does not constitute financial advice. Pure Benchmarks is a benchmarking and data platform.