AI Can Analyse Your Investments, But Should You Trust It With Your Money?

Outside his role as Chief Product Officer at aviation strategy firm SimpliFlying, Shubhodeep Pal enjoys building vibe-coded, AI-powered apps for himself. When the investor in him wanted to “take the whimsy out” of his investment strategies, he turned to AI. “AI built the system within days under my guidance, complete with data, signals, even a back-test going as far as 2018,” he recalls. “On the first run with real market data, it caught and fixed five of its own bugs.”

How then does he assess AI as an investment adviser? “It can be very wrong while sounding very wise,” he admits. His rule is: “It proposes, but I approve every trade.” For Pal, any result that looks too good is a bug until proven otherwise. “No one, not even AI, can predict markets,” he says. “The most exciting part is that systematic investing, which would ordinarily need a huge quant team or an expensive degree, can now be run from a laptop.”

Clocking In With AI

Pal’s approach holds a vital lesson for a growing number of retail Indian investors seeking investment advice from AI—the newest oracle in town that comes dressed in an algorithm, talks with near certainty about a notoriously uncertain space and loses no sleep over the Street’s tantrums despite having no qualifications or experience. What makes the lesson even more pertinent is that these investors, unlike Pal, are using generic tools such as ChatGPT, Claude and Gemini.

Considering how integral these tools have already become to everyday life in India, their reach across the country’s direct equity investors—with more than 22.9 crore demat accounts as of May 2026 and a median age of 32—is anybody’s guess. After all, AI tools listen without judging and are always available. It should be music to their ears that a recent analysis of AI’s investment advice by researchers from Massachusetts Institute of Technology and Stanford University has inferred that AI “aligns remarkably closely with expert advice.” Yet, as with all things AI, caution ought to be the buzzword. The same study has also revealed that the outcome depends heavily on how the AI is queried.

For example, unlike men who used terms like “growth,” “crypto” or “strategy” a lot more in their questions, women, who often used words related to everyday life, were recommended a lower-risk strategy with a smaller allocation of stocks, resulting in assets that were, on average, about five percent lower. Even with identical questions, the AI tool recommended more cautious strategies when it perceived the user to be a woman.

Research Sidekick

“AI tools can aid research but financial advice is a personal matter and without understanding the user’s financial profile and risks, generic tools cannot provide meaningful guidance,” says Rohit Prakash, Founder of Genvest, a SEBI-registered AI wealth management platform. “On the contrary, regulated AI platforms like ours analyse income, savings, portfolio, age, risk profile and even market conditions before recommending asset allocation, fund selection and rebalancing. This is where such platforms can make a real difference, especially in India where there are fewer than 1,000 registered investment advisors (RIAs) for more than 20 crore investors.”

Rohit avers that the debate should not be human versus AI. “AI can significantly enhance productivity by analysing large volumes of data and identifying patterns,” he says. “It can also go deeper and offer personalisation.” According to Rohit, its true value lies in continuous, real-time monitoring — evaluating whether a new stock or fund is adding to the existing portfolio, whether the risk is worth taking and, more importantly, alerting the investor when it is the right time to exit.

“Modern AI tools are good at analysing investments, but cannot be relied upon for big-ticket decisions,” concurs Akhil Theerthala, a senior data scientist and independent researcher working on AI evaluation and reliability. “Their recommendations can readily change even with minor changes in prompts or conversational contexts. They can throw up weak or fabricated citations. There is no practical way yet for tracing their outputs to the underlying training data, which limits our ability to verify their rationale. Besides, they don’t always tell you when they are wrong or unsure, which is unacceptable when the stakes are high.”

Kanan Bahl, a chartered accountant, personal finance educator and Founder of Fingrowth Media, recommends taking advice from RIAs. “Investment advice is supposed to be extremely personal as each person’s goals and their ability and willingness to take risks differ,” he says. “AI cannot understand that and hand-hold you during your 20-30-year journey of earning and investing money.”

Human Edge

The limitations of generic AI tools make human judgement even more critical. “The tool may lack important context about the investor’s changing needs, existing portfolio and financial position, and should not be treated as an unquestionable source of financial advice,” says Rohit. “That is why Genvest has adopted a human-in-the-loop approach.”

Theerthala points out that the financial benchmarks used to test these tools still rely on simplified tasks and information taken from textbooks, exams or question-and-answer sets. “In the case of open-ended investment advice, a lot is riding on getting the facts right, understanding the user’s situation, being aware of the risks involved and staying consistent,” he says. “These tools cannot handle all that very well. So while they are helpful for early-stage research, they cannot yet be fully trusted for investment advice.” In marrying artificial intelligence with real money, being smart is about being sceptical, counting more on questions than answers to chart the safest path ahead.

Don’t treat AI as a financial adviser; it is best used as a research tool.

Don’t blindly follow AI’s recommendations.

Don’t expect AI to predict markets.

Don’t trust unusually good-looking returns.

Don’t ignore your own risk tolerance.

Don’t share sensitive personal and financial information.

Don’t rely on outdated or inaccurate data.

Don’t confuse a plausible answer with a sound one; confidence is not the same as competence.

Published on: Sunday, August 23, 2026, 07:20 AM IST