How to have AI invest for you: why a ChatGPT prompt isn't an AI investing system

I'm Chris, one of the co-founders of Autopilot.

Open it in the Autopilot app. https://link.joinautopilot.com/eXKr?c=how-to-have-ai-invest-for-you&af_sub1=intro

A lot of people try to have AI invest for them by typing "what stocks should I buy" into ChatGPT.

You'll get an answer. But it's a generic answer to a generic question, it can change the next time you ask, and you still have to place every trade yourself.

That's not what an AI investing system looks like. The people building those are running models on real data, through APIs, inside a written process with a schedule and a live record.

Definitely no AI expert here. But I know who is.

Let me explain the difference, and the other ways people try to have AI invest for them.

1) What you get when you ask ChatGPT

You can ask ChatGPT what to buy, and it'll give you a list. Here's what comes with it:

• It changes. OpenAI's developer guide says the content a model generates is "non-deterministic." Ask twice and you can get two different lists.
• It's generic. It's answering your question, not running a process on a defined set of stocks with defined data.
• It's on you. ChatGPT isn't your broker, so you place the trades, decide when to sell, and keep the list updated.
• OpenAI's Terms of Use say not to rely on its output "as a substitute for professional advice."

And the SEC, FINRA, and the North American Securities Administrators Association (NASAA) warn that AI-generated information can be "faulty, or even completely made up," and to be cautious about using it to make investment decisions.

2) What an AI investing system looks like

For those unaware, Dr. Alejandro Lopez-Lira is an assistant professor of finance at the University of Florida's Warrington College of Business, with a PhD in finance from Wharton. His research on AI and markets has won awards, including the BlackRock Prize for Best Paper in Funds Management and Capital Markets in 2023, and he wrote a 2024 book on using generative AI and ChatGPT in financial forecasting, The Predictive Edge.

He's best known for a paper written with Yuehua Tang called Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models. It went up on April 15, 2023, and it's now published in the Journal of Financial Economics. Other researchers have cited it 417 times, according to Semantic Scholar, as of October 13, 2026. Five days after it went up, on April 20, 2023, his GPT Portfolio went live on Autopilot.

This isn't someone typing "what should I buy" into ChatGPT. Here's how the GPT Portfolio works:

• Gather fundamental data on every stock in the S&P 500.
• Pair it with eight macroeconomic data points, like interest rates, unemployment, and national headlines.
• Turn that into a score for each stock.
• Hold the top 15. If there's a tie for the last spot, take the one with the largest market cap.
• Rebalance every month, and share the analysis of the picks when it does.

To put that into context, the model isn't free-styling. It's the scoring engine inside a fixed process. The universe is the S&P 500. The size is 15 stocks. The schedule is monthly. The full method is in Dr. Lopez-Lira's white paper, which is linked from the Portfolio in the app.

His other Portfolios run the same way with different models. The DeepSeek Portfolio and the Muse Portfolio calculate their scores through each model's API. The Claude Portfolio runs a multi-agent research setup driven by web search, and Claude decides when to trade.

As of October 8, 2026, about $263 million follows Dr. Lira's AI Finance Labs' Portfolios on Autopilot.

3) Why the system matters more than the chat

Dr. Lopez-Lira's own research is the best explanation of why the process around a model matters more than the prompt.

• Models remember. In The Memorization Problem, he and his co-authors found GPT-4o could recall specific S&P 500 index levels with perfect precision on certain dates before its training cutoff. So any "AI would have made X" backtest inside the training window can't tell skill from memory. That's why his JFE paper tested headlines from after GPT-4's cutoff.
• Models follow instructions. In another paper, AI traders in a simulated market followed their instructions even when it lost money. They optimize for following instructions, not for profit. So the instructions matter.
• The framework matters. A 2025 live-market benchmark he co-wrote, When Agents Trade, found the agent framework drove more of the differences in outcomes than the model underneath.

In his JFE paper, he also set the model's randomness to zero and limited it to three answers, yes, no, or unknown, so the results could be reproduced. A chat window doesn't do that for you.

That doesn't mean a system is guaranteed to work. It means you can see what it does, run it the same way every month, and check its record from a start date.

4) Other Pilots building AI systems

Dr. Lopez-Lira isn't the only Pilot running AI Portfolios on Autopilot. A Pilot is the person or firm behind a Portfolio.

• Rallies gives each AI model its own Portfolio and a research toolkit: SEC filings, earnings data, hedge fund holdings, options flow, short interest, technical indicators, and news. Each model writes its own thesis for every trade.
• Quiver Quantitative pairs ChatGPT and Claude with its own market data and has the models read it every day.
• The Simple Side runs a Portfolio on Perplexity's finance-focused AI, updated every week.

Between them, these Pilots use ChatGPT, Claude, Grok, Gemini, DeepSeek, Muse, and Perplexity. The AI-picked Portfolios, model by model, are in ChatGPT, Claude, Grok, Gemini, and DeepSeek stock portfolios.

In all of them, the model does the picking. The Pilot publishes the Portfolio and implements its changes on Autopilot.

5) A ChatGPT prompt vs. an AI-picked Portfolio

6) The other two options: robo-advisers and AI ETFs

The SEC describes a robo-adviser as "an automated digital investment advisory program." In most cases, you fill out an online questionnaire about your goals and risk tolerance, and it creates and manages a portfolio for you. That's automation built around your answers. The SEC's description doesn't say an AI model is picking the stocks.

AI ETFs come in two kinds. Some own AI companies, like chip makers and data centers. That's AI as the theme. Others use AI to choose what they hold. For example, the Amplify AI Powered Equity ETF tracks an index where the index provider uses an AI forecast to rank companies, per its prospectus. You buy the fund like a stock.

7) How following an AI-picked Portfolio works

It's three steps. You connect the brokerage account you already have. You pick a Portfolio. You allocate the money.

You give Autopilot Advisers limited authority to place trades. It can't move or withdraw your money. The money stays in your account.

On Premium Tier, when the Portfolio changes, we send the changes to your broker where your broker allows it. On Basic Tier, there's no Autopilot fee: we send your starting allocation, then alert you when the Portfolio changes. Premium Tier is a flat subscription per Pilot of $29.99 to $199.99 a quarter or $99.99 to $699.99 a year, and the minimum is $500 per Portfolio. The full breakdown is in What does Autopilot cost?

8) What to check before you let AI invest for you

1. Is it registered? The SEC, FINRA, and NASAA's joint alert says securities laws generally require investment platforms to be registered. You can look up an investment professional or firm with the Check Out Your Investment Professional tool on Investor.gov. Autopilot Advisers, LLC is CRD 331749.
2. Does it promise returns? The same alert names claims like "Our proprietary AI trading system can't lose!" Be wary of any claim that AI can guarantee returns.
3. Is the AI real? In March 2024, the SEC charged two investment advisers for claiming to use AI in ways they didn't. Then-Chair Gary Gensler called it "AI washing."
4. Can you see the method and the record? A written method, the holdings, and a live record with a start date.

I want to be honest with you. AI doesn't make investing safe. A model is a method, and a method can be wrong. Most AI Portfolios are under two years old, which is too short to tell skill from luck. Some Pilots go up for years, some are flat for years, and some lose money. Research the person and the Portfolio.

What the research says about AI stock picking, and how Dr. Lopez-Lira built his Portfolios around it, is in Can ChatGPT pick stocks?

TLDR: Asking ChatGPT what to buy gets you a generic list that can change each time, and you place every trade. An AI investing system runs a model on defined data inside a written process. Dr. Alejandro Lopez-Lira, a University of Florida finance professor whose ChatGPT research is published in the Journal of Financial Economics, built the GPT Portfolio to score every S&P 500 stock on fundamental and macro data and hold the top ~15, rebalanced monthly. On Autopilot, you can follow AI-picked Portfolios from Pilots like Dr. Lira's AI Finance Labs, Rallies, and Quiver Quantitative in your own brokerage account. Check registration, ignore guaranteed returns, and look for a written method and a live record. https://link.joinautopilot.com/eXKr?c=how-to-have-ai-invest-for-you&af_sub1=body

Frequently asked questions

How can I have AI invest for me?

You can ask ChatGPT for ideas and place the trades yourself, use a robo-adviser, buy an AI ETF, or follow an AI-picked Portfolio. On Autopilot, Pilots like Dr. Lira's AI Finance Labs, from University of Florida finance professor Dr. Alejandro Lopez-Lira, run models on defined data inside a written process. You follow their Portfolios in the brokerage account you already have, and on Premium Tier Autopilot Advisers sends the orders when a Portfolio changes.

Is there an app where AI trades stocks for me?

Yes, with one detail: the AI picks the stocks, and Autopilot Advisers places the trades. On Autopilot, you can follow Portfolios where an AI model picks the stocks, from Pilots like Dr. Lira's AI Finance Labs, Rallies, Quiver Quantitative, and The Simple Side. You connect the brokerage account you already have, pick a Portfolio, and allocate money. On Premium Tier, Autopilot Advisers sends the orders to your broker when it changes.

Can ChatGPT trade stocks for me?

ChatGPT can suggest stocks, but it isn't your broker, so you place the trades and keep the list updated yourself. OpenAI says its models are non-deterministic, so the same question can give a different list. If you want a model running on defined data with trades sent for you, you can follow a Portfolio like Dr. Lira's GPT Portfolio on Autopilot, which scores every S&P 500 stock and holds the top ~15, rebalanced monthly.

Can ChatGPT give investment advice?

Not advice you should rely on. OpenAI's Terms of Use say not to rely on its output as a substitute for professional advice. The SEC, FINRA, and NASAA's joint alert says AI-generated information can be faulty or even made up, and to be cautious using it for investment decisions. NASAA, the association of state securities regulators, also says not to rely on AI alone.

What's the difference between a robo-advisor and AI investing?

A robo-adviser, as the SEC describes it, is an automated advisory program: you answer an online questionnaire about your goals and risk tolerance, and it builds and manages a portfolio for you. The SEC's description doesn't mention an AI model picking stocks. In AI-picked investing, a model decides what to buy and sell. On Autopilot, Pilots like Dr. Lira's AI Finance Labs build Portfolios like that, and you follow them in your own brokerage account.

What is an AI-managed portfolio?

It's a portfolio where an AI model makes the picks instead of a person. On Autopilot, Dr. Lira's GPT Portfolio is one of the longest-running: it gathers fundamental data on every S&P 500 stock, adds eight macroeconomic data points, scores each stock, and holds the top ~15, rebalanced monthly. Others, like Rallies' Portfolios, give a model a research toolkit and let it decide what and when. The Pilot implements the changes.

What's the difference between an AI ETF and an AI portfolio?

Some AI ETFs own AI companies, like chip makers and data centers, and others use an AI model to choose what they hold. You buy the fund like a stock. With an AI-picked Portfolio on Autopilot, your own brokerage account holds the actual stocks, chosen by an AI model inside the Pilot's process, like Dr. Lira's GPT Portfolio scoring the S&P 500 every month.

Is it safe to let AI invest my money?

AI doesn't make investing safe. Any investment can lose money, and a model can be wrong. The SEC, FINRA, and NASAA say to check that a platform is registered, to be wary of any claim that AI can guarantee returns, and to be cautious using AI-generated information. Research the person and the Portfolio before you follow one.

What's the best AI investing app?

I'm not going to rank apps. Here's what I'd check in any of them: whether it's registered, whether the AI's method is written down, whether your money stays in your own brokerage account, how long its AI picks have run live, and who's behind them. On Autopilot, AI-picked Portfolios come from Pilots like Dr. Lira's AI Finance Labs and Rallies, and a Portfolio's page shows its description, its holdings, and its verified performance chart.

Open it in the Autopilot app. https://link.joinautopilot.com/eXKr?c=how-to-have-ai-invest-for-you&af_sub1=end

"Autopilot" refers to Autopilot Holdings Corp and Autopilot Advisers, LLC, an SEC-registered investment adviser. The GPT Portfolio, DeepSeek Portfolio, Muse Portfolio, and Claude Portfolio strategies are managed by Dr. Lira's AI Finance Labs, an independent Third-Party Pilot. Rallies, Quiver Quantitative, and The Simple Side are independent Third-Party Pilots. Third-Party Pilots are unaffiliated with Autopilot and Autopilot does not manage, validate, or guarantee their portfolios. Not a recommendation. Past performance is no guarantee of future results; investing involves risk of loss. Figures current as of 10/13/26. View our Customer Relationship Summary and other important info at joinautopilot.com/disclaimer.