At a time when more and more savers are consulting ChatGPT before making a financial decision, should we really trust it? Between a promising educational tool and a source of incomplete answers, AI is intriguing, but remains far from replacing an advisor.

Should you trust ChatGPT to manage your savings? Recently, the French Financial Markets Authority (AMF) issued a warning: these tools, however sophisticated they may be, "do not replace investment advice from a financial professional."

However, it's undeniable that for over a year now, artificial intelligence has been making its way into financial conversations: some use it to understand savings products, others to get a quick opinion on a retirement savings plan (PER) , life insurance , or the tax implications of an investment. But beyond the hype, a question remains: how far can we rely on AI to manage our money?

“It’s tempting to believe that a robo-advisor or virtual agent will be able to answer all of an investor’s questions perfectly,” says Figen AI, a company dedicated to wealth management and finance professionals that offers a suite of specialized AI assistants. “However, the illusion of infallible intelligence quickly clashes with reality. First, no AI can predict market movements with certainty or fully grasp the nuances of each individual situation.”

Information that is often incomplete

According to Aymeric Richard, director of the firm Chartrons Patrimoine, its use remains very limited among his clients. He explains that while ChatGPT " allows for the collection of some information, this tool remains incomplete . If you ask the AI ​​for information about life insurance, for example, it might tell you in detail what life insurance is, but not what role it can play in your assets and whether or not it might be a good solution for your specific situation."

He acknowledges, however, that AI is a "good tool for having a more in-depth conversation with a client," as it can address the financial literacy gap among most savers. "At school, no one learns how to manage their budget or savings; it's not innate. AI is interesting because it will provide in-depth access to knowledge about the principles of wealth management, much like the internet. For wealth management advisors, being with a knowledgeable saver could ultimately save time and allow them to offer even more comprehensive solutions."

"My clients sometimes arrive with poorly understood concepts or biased reasoning."

At Paul Ruben Brami's firm, 26 Patrimoine, the observation is different. " Many of my less tech-savvy clients do indeed start by asking ChatGPT," he notes. "They then arrive with poorly understood concepts or biased reasoning. The AI ​​provides information, but it's never perfectly tailored to the client's situation."

The problem often stems from a lack of context provided to the tool: "For ChatGPT to respond correctly, you would need to tell it your entire personal, professional, and family situation. Customers don't do this, or do it incorrectly, so the response isn't accurate. "

These limitations are not new. Conversational AI has long been known for providing inaccurate or outdated information. While their models are becoming more sophisticated, advisors remain very cautious . "Yes, it's evolving," admits Aymeric Richard. "But not everything is usable yet. We're at the beginning. In some areas, we're already seeing progress, but it will still be a few years before we have a perfectly reliable tool."

Information not always up to date

For him, the risks lie both in the accuracy of the answers and in how users employ them: "We must be careful about what AI can tell us. It doesn't necessarily answer everything, and it can lead us down the wrong path. We mustn't use it indiscriminately, nor replace our ability to think ."

Paul Ruben Brami, for his part, points to another danger: obsolescence. “AI is never truly up-to-date. Between political decisions, budget allocations, tax measures… a single announcement is enough to change everything. AI can’t keep up with that pace.” He cites the example of a technical question related to the PER (Retirement Savings Plan): “I received completely contradictory answers compared to what the official documents indicated. As soon as you get into very specific details, AI isn’t reliable. Even I don’t trust it on certain specific concepts.” For him, the conclusion is clear: “It’s good for learning, but not for making decisions. It’s not enough.”

However, both advisors acknowledge that AI has real potential, provided it's used for what it's designed to do. "It's very useful for simplifying concepts," admits Paul Ruben Brami. "If an article is too complex, asking AI to simplify it can help a client grasp the essentials. That's invaluable." He also uses it in his own work, but only occasionally: writing emails, rewording, preparing educational materials. "Never for allocating funds, never for providing advice."

Aymeric Richard shares this vision. For him, AI is a "time-saving tool," not so much in consulting itself, but in peripheral tasks: writing, organizing, and researching. He also sees it as a way to support a transformation of the profession. "We can imagine, within five years, AI tools being made available to clients to help them go further, and us, as consultants, retrieving the information to improve communication. This could enrich the relationship rather than replace it."

Far from doomsday scenarios or promises of instant wealth, professionals are therefore calling for its judicious use. AI can inform, simplify, and structure an idea. It can help a saver better understand what a retirement savings plan (PER) is, distinguish a euro fund from an ETF, or grasp the basics of taxation. It can even help take the plunge and schedule a meeting with a financial advisor, by sparking a desire to learn more. But it remains blind to your personal situation, which is sometimes outdated, often imprecise on subtle concepts—and incapable of integrating the human, tax, and legal aspects of a wealth management plan.

In short, AI can be an excellent starting point. But certainly not an end point. As Paul Ruben Brami sums it up, "it's good for education, but not for decision-making." It's then up to each individual to determine where to draw the line between digital autonomy and human guidance.