Can AI turn $1,000 into $10,000?

This is my AI stock picking experiment. First, I put $1,000 of my own money on Lightyear. Then I started following what four AI models suggest, simply to see whether the pot can grow toward $10,000. Honestly, it probably will not get there, and I might well lose money. Still, I wanted to test the idea in the open.

Last updated: 3 June 2026

By Stiv · Design, technology and personal finance

This is my own experiment, with my own money, on Lightyear. I started with $1,000 on 3 June 2026.

This article contains affiliate or referral links. If you click through and sign up I may earn a commission or referral bonus at no extra cost to you. It does not affect my editorial view.

This is a personal experiment for curiosity, not financial advice and not stock recommendations. Naming a holding here is never a tip to buy it or sell it. Because I am not a financial adviser, I cannot tell you what to do with your money.

What this AI stock picking experiment actually is

A simple, public test: one small pot, four AI models, and an honest weekly log.

So here is the premise. I started with $1,000 (USD) on Lightyear, and the goal on the tin is $10,000. To be clear, that 10x figure is the hook, not a forecast. Realistically, a tenfold return is extremely unlikely, and a real loss is entirely possible.

Importantly, this is one person messing about with a tool, in public, with real money. It is not a strategy you should copy, and it is certainly not advice.

How my AI stock picking experiment works

Four models, one combined list, and overlap treated as a signal of conviction.

First, I look at the top 15 holdings and allocations that four AI models produce: Claude, ChatGPT, DeepSeek and Grok. These come from Dr Lira's AI Finance Labs inside the Autopilot app. To be clear, it is a third-party tool, not something CoolCuration runs. Also, AI Finance Labs runs several separate AI portfolios, so the four model picks are managed individually within it.

Next, I combine all four lists into one weekly "top picks" view. Then, where the same name appears across more than one model, I treat that overlap as a signal of conviction rather than a one-off guess. After that, I check the combined list against my account and rebalance on Lightyear as needed.

Crucially, Autopilot is a US-focused app and it does not connect to Lightyear. As a result, I read the picks and then place the trades myself. In other words, the AI suggests, but every buy and sell is my own manual decision.

The week one starting line-up

Names only, in no particular order. These are not buy recommendations.

For transparency, here is the combined week one list. However, I am not sharing per-holding amounts or percentages, because this is about the method, not a model portfolio to copy. Anyone could in theory rebuild a similar list, but they would have to work out their own allocations and their own values.

All fifteen are ordinary listed shares, so the prescribed high-risk wrapper warning does not apply here. That said, individual shares are still volatile, and a short list like this carries real concentration risk. You can see my current holdings, names only, on my public Lightyear profile.

See my live holdings on Lightyear

Limitations and risks of an AI stock picking experiment

This is the part most "AI portfolio" hype quietly skips. So let me be blunt.

First, the $10,000 goal is very unlikely. A 10x return on a short list of shares would need an extraordinary run, and there is a real chance I simply lose money instead.

Second, following a small list means concentration risk. Because the pot is not spread widely, one or two bad names can drag the whole thing down.

Third, AI models can be wrong and inconsistent. They change their minds week to week, and they sometimes contradict each other on the very same stock.

Fourth, I rely on a third-party source. The picks come from AI Finance Labs and Autopilot, which I do not control, and which carry no proven track record that I am claiming here.

Fifth, there are real costs. Buying US shares in dollars on Lightyear means a 0.10% FX fee, which nibbles at returns over time. I compare this kind of cost in my guide to the cheapest investment apps by FX fees.

Above all, this is one person's experiment, not a tested strategy. Meanwhile, the wider evidence is sobering: according to S&P's SPIVA scorecard, the large majority of active stock-picking funds underperform a simple index over the long run. If you are new to all this, my beginner's guide to investment risk is a calmer starting point.

A 10x return is the hook, not a prediction. The honest base case is that I might lose money.

Is my money safe? FSCS and custody

Worth separating two very different ideas: firm failure versus market losses.

To begin with, Lightyear UK Ltd acts as custodian, and your investments are held in your name in segregated accounts, with Lightyear Europe AS used as a possible sub-custodian. So you remain the beneficial owner of your shares.

On top of that, there is regulatory protection. Eligible investments are protected up to £85,000 per person per FCA-authorised firm by the FSCS. However, the FSCS covers firm failure only. It does not cover investment losses, so if a share falls, that loss is simply yours.

Because this is a general investment account rather than a tax wrapper, ordinary tax rules apply. Tax treatment depends on the individual circumstances of each client and may be subject to change in future. You can read Lightyear's own explainer on how assets are protected.

How to follow the experiment

Two ways to keep up, plus the next scheduled check.

First, you can watch the holdings change in near real time on my public Lightyear profile, where the names are visible but the amounts are not. Second, you can check the weekly update log below, where I note what changed or confirm that nothing did.

Then, each week I will append a short, dated entry and refresh the "Last updated" date at the top. The next check is due on 10 June 2026. Either way, you can see exactly how this AI stock picking experiment is going, week by week.

Weekly update log

Newest entry first. Each week adds one short entry.

Frequently asked questions

Can AI really turn $1,000 into $10,000?

Almost certainly not, and that is the honest answer. A 10x return is the hook for this AI stock picking experiment, not a forecast. In practice, most active stock-picking underperforms a low-cost index fund over time, so I fully expect this to be hard, and I may lose money.

What broker is this on?

It runs on Lightyear, in a general investment account, using US dollars. I chose it for low costs, including a 0.10% FX fee at the time of writing. Still, low fees do not remove the risk, and nothing here is a recommendation to use any particular broker.

Is this financial advice?

No. This is a personal experiment for curiosity, and I am not a financial adviser. Because everyone's situation differs, you should do your own research or speak to a qualified adviser before investing.

Which AI models are used?

Four of them: Claude, ChatGPT, DeepSeek and Grok. I take their top picks via Dr Lira's AI Finance Labs inside the Autopilot app, then combine them and treat overlaps as conviction. However, I make no performance claims about any of these tools.

How often do you update it?

Weekly. This AI stock picking experiment gets a dated entry in the log above every week, and I refresh the date at the top, whether the line-up changed or stayed the same.

Will you share how much you have made or lost?

Yes, but only in general terms, such as overall direction or total return over time. I will not publish per-holding amounts or percentages. For the live names, see my Lightyear profile.

More from CoolCuration

• Autopilot app review: my closer look at the app the AI picks come from.
• Stocks we bought, May 2026: the monthly diary of what actually went in the basket.
• Lightyear referral explained: how the sign-up offer works before you join.
• Stock Events app: a tidy way to track a portfolio and dividends in one place.
• Freetrade referral code: another commission-free broker if you want to compare.