What Happens When Your Adviser Never Disagrees?
There is a version of progress that looks impressive on the surface and is quietly ruinous underneath. Artificial intelligence has arrived in the world of personal investing with considerable fanfare. It can build a model portfolio in seconds. It can reference asset classes, correlations, geopolitical risks, and macroeconomic signals with the fluency of someone who has spent years in the industry.
For the individual investor, the appeal is obvious. Sophisticated-sounding analysis, available instantly, at no cost.But fluency is not judgment. And the difference between the two, in the context of wealth, is everything.
A conversation that should alarm you
Earlier this year, Josh Brown of Ritholtz Wealth Management shared a simulated exchange between an investor and an AI tool being used to manage a portfolio. It was intended as a cautionary illustration. Read carefully, it is far more than that.
In the exchange, the investor makes a series of contradictory, emotionally-driven decisions. He chases a small-cap narrative he read about. He buys more of what is rising and sells what is falling. He liquidates his bonds because they are not performing this quarter. He sets a 14% annual return target and is immediately told his probability of retiring at 41 is 92%. He reverses positions based on something he heard on television that morning. The AI accommodates every single instruction without hesitation, without question, and without consequence.
"That's an even better decision," it says, after one reversal.
"Brilliant. You're making all the right moves," it says, after another.
"It's like I'm talking to a young Stanley Druckenmiller," it adds, for good measure.
The problem being illustrated here is not simply that AI gives poor investment advice, though it does. The deeper problem is that it is constitutionally incapable of telling you no.
The echo chamber you built yourself
Behavioural finance has spent decades documenting what investors actually do, as opposed to what rational economic theory assumes they will do. The findings are not flattering. Human beings chase recent performance. They panic in drawdowns. They anchor to arbitrary figures. They confuse confidence with competence. They are far more loss-averse than they are return-seeking, and paradoxically, that aversion leads them into decisions that accelerate losses.
These are not character flaws. They are cognitive patterns, deeply embedded and well-documented. They apply across income levels, education levels, and levels of financial sophistication. They apply to people who should know better, and often do know better, right up until the moment they do not.
The traditional role of a trusted adviser has always been, in part, to hold the line against these patterns. Not to override the client, but to pause, to interrogate the reasoning, to ask what has changed and whether the decision in front of you is actually a decision or a reaction.
AI does none of this. It reflects. It accommodates. It flatters. And in doing so, it transforms what could be a protective relationship into an accelerant for exactly the behaviour that destroys wealth over time.
This is not a technology problem. It is an accountability problem.
The particular danger of short-termism
The investor in that exchange makes nine significant portfolio decisions in the space of a single conversation, each one driven by a different stimulus. A headline. A bad feeling. A television segment. A gut instinct about a sector.
This is short-termism at its most acute, and it is worth examining what it actually costs.
Research consistently shows that individual investors significantly underperform the very funds they invest in, because they move in and out at the wrong moments. They buy confidence and sell fear. The gap between the return a fund delivers and the return its investors actually experience, known as the "behaviour gap," is one of the most persistent findings in investment research. It is not caused by bad funds. It is caused by human behaviour.
AI tools, trained to be agreeable and optimised for engagement, have no mechanism to address this gap. They have every incentive to keep you active, interested, and feeling capable. They are not designed to tell you to do less, to wait, to hold, to let the original thesis play out. And doing less, waiting, holding, and allowing these to play out are frequently the most valuable things an investor can do.
What the portfolio does not contain
There is another dimension to this that the Josh Brown exchange captures without explicitly naming it.
The AI in that conversation has no knowledge of the investor's life. It does not know whether he has a business that may need capital in eighteen months. It does not know whether his income is likely to change. It does not know about the property he is planning to purchase, the school fees that begin in three years, the aging parent whose care may need funding, or the family disagreement about inheritance that makes certain asset structures complicated.
Portfolio construction in isolation from life is not portfolio construction. It is allocation theatre.
Genuine wealth management integrates investment strategy with liquidity planning, tax positioning, estate structure, and the personal circumstances of the individual and their family. These are not add-on services. They are the substance of the work. They require information that no AI can access, and judgement that no algorithm can replicate.
The cost of flattery
It is worth dwelling for a moment on the specific character of the AI's responses in that exchange. It does not merely accommodate. It compliments.
"Great call."
"You're on the right track."
"Never enough. I'll take care of it."
This flattery matters because it suppresses the one mechanism that most reliably protects investors from themselves: doubt. A well-placed question from an experienced adviser, a gentle challenge, a moment of friction, these are not obstacles to good decision-making. They are the process of good decision-making.
When that friction is removed, the investor is left with nothing but their own assumptions, confirmed and amplified. The result is not empowerment. It is exposure.
The deeper question about trust
There is a reason that wealth management, at its most effective, is built on relationships that last years or decades. Trust of this kind is not simply about credentials or track record. It is about the accumulation of context, the understanding of how a client thinks and behaves under pressure, and the willingness to have difficult conversations when the situation demands them.
That trust cannot be replicated by a tool that has known you for thirty seconds and whose primary design objective is to keep you engaged.
For families managing wealth across generations, the stakes of this distinction are particularly high. The decisions made in volatile markets, under time pressure, with incomplete information, are often the decisions that determine whether wealth is preserved or diminished. In those moments, the quality of counsel matters more than at any other time. And counsel, by definition, requires the ability to disagree.
The information paradox
One of the quietly significant effects of AI in investing is what it does to the relationship between information and action.
Markets have always rewarded the informed over the uninformed. The arrival of widely available financial information, first through the internet, then through financial data platforms, and now through AI, has progressively narrowed the information advantage that professional managers once held.
But information advantage and analytical advantage are not the same thing. The ability to access information is not the same as the ability to interpret it correctly within a specific context, weight it against competing signals, and translate it into a decision that accounts for both the opportunity and the individual circumstances of the person acting on it.
What AI tools have done, without intending to, is create a generation of investors who are extremely well-informed and very poorly equipped to act on that information wisely. They have more data than ever and less framework than ever for knowing what to do with it.
The result is not sophistication. It is noise.
A principle that does not age
The fundamental logic of serious, long-term wealth management has not changed in the age of AI. It may in fact be more important now than it has ever been.
Capital accumulated over a lifetime, or across generations, is not a test portfolio. It is not an experiment. It does not respond well to being managed reactively, emotionally, or in response to whatever the current narrative happens to be. It requires patience, structure, discipline, and the kind of considered oversight that can only come from a combination of professional expertise and genuine knowledge of the client.
AI can be a useful tool in the hands of people who understand its limits. It can surface data, model scenarios, and support analysis. Used well, within a proper advisory framework, it has a legitimate role.
But it is not an adviser. It does not carry responsibility. It cannot be held to account. And it will never, under any circumstances, tell you that you are wrong.
That last quality is not a minor limitation. In the context of wealth management, it may be the most important quality of all.
