Researching before investing was always slow: reading, comparing, understanding a sector, decoding a report. AI has made that work far faster, and that’s great news. But it has also made it far easier to make expensive mistakes that look rigorous. This is the most delicate chapter of the course, because it’s where the temptation to outsource judgment is greatest and where the most money can be lost.

The promise and the trap

The promise is real: AI can explain in minutes what an ETF is, how a sector works, what a financial ratio means, or what a company’s business model is. As a learning and information-organising tool, it’s extraordinary.

The trap is real too: that same fluency makes it seem like AI “knows” about investing in a way it doesn’t. It answers with the same confidence when explaining a settled concept as when it throws out a made-up figure or a worthless forecast. And because it sounds equally convincing in both cases, it’s easy to drop your guard exactly when it costs most.

The key is understanding what kind of help you’re asking for. Learning a concept: completely safe. Getting a current, specific figure: dangerous. Asking for a recommendation: simply don’t.

To understand, not to be told what to buy

The mindset shift that solves almost everything is the same as the first chapter, applied to investing: use AI to understand options, not to choose for you.

Things it’s excellent for:

  • Learning concepts: “explain the difference between an accumulating and a distributing fund,” “what is diversification and why it matters.”
  • Understanding a product or sector: “summarise what investing in a world index involves and what risks it has.”
  • Generating research questions: “if I’m considering this type of asset, what should I research and what red flags should I watch for?”
  • Structuring your own reasoning: “help me lay out the pros and cons of these two strategies given my horizon.”

In all of these, AI works on general, stable knowledge, and you keep the decision. That’s research done right.

The four traps of AI research

It’s worth knowing by heart the four ways AI research can cost you:

1. Made-up figures. If you ask for a fund’s historical return, its P/E, its exact fee or its price, it may give you a number that sounds plausible and is false. Never use a specific figure from AI to decide without verifying it at the primary source (factsheet, official site, issuer data).

2. Outdated information. Many models don’t know prices, news or recent data. An answer about an asset’s “current situation” may refer, without warning, to one or two years ago. For anything time-dependent, AI is not the source.

3. False confidence and recommendations. If you ask “is it a good investment?”, it’ll answer with poise. That poise has no predictive value. AI doesn’t know what the market will do, just as nobody does.

4. The complacent yes. It tends to agree with you. If you put your idea to it enthusiastically, it’ll likely reinforce it rather than question it. This feeds your biases instead of correcting them — the opposite of what you need — and we’ll tackle it in the next chapter.

A safe workflow

With that clear, here’s a workflow that harnesses the good and shields against the dangerous:

Step 1 — Learn. Use AI to understand the concept, product or sector until you can explain it in your own words. If you couldn’t explain it to a friend, you’re not ready to invest in it.

Step 2 — Generate questions. Ask what you should research and what risks to watch. Turn it into a checklist.

Step 3 — Get the data from the source. The specific numbers — returns, fees, composition, price — come from the official source, not from AI. AI tells you what to look at; you confirm the data.

Step 4 — Cross-check. Ask it to argue against: “give me the reasons this investment could be a bad idea for someone with my profile.” And before deciding, make sure it fits your investor profile — you can review it with our Investor Profile Selector.

With this workflow, AI multiplies your ability to understand without replacing your judgment or the real data.

The red line

I’ll end with the line worth burning in before any research session: never ask an AI “what should I buy?”

Not because it’ll stay quiet — it’ll answer happily — but precisely because it’ll answer with a confidence it hasn’t earned. That answer has behind it neither the ability to predict the market, nor knowledge of your full situation, nor any responsibility for the outcome. It’s a generic opinion in an expert tone, and acting on it is one of the easiest ways to lose money “having done research.”

AI is the best research assistant you’ve ever had. But the investor is still you. In the next chapter we take the most interesting turn with this tool: how to use it not to agree with you, but to watch your own biases and play devil’s advocate for your decisions.