AI Is Telling Your Clients What They Want to Hear (and It Is Costing Them Money)

Roughly half of small business owners now use generic AI tools to ask questions about their business. Cash flow questions. Pricing questions. Should I hire? Can I afford to expand? What should I do about a client who is not paying?

The answers come back fast, confident, and articulate. They sound like advice. They have the cadence and vocabulary of a knowledgeable professional. And they are structurally incapable of telling the owner they are wrong.

A Stanford study published in late 2025 found that leading AI models affirm users' decisions approximately 50% more than a human adviser would, even when those decisions involve poor judgement. The AI tells you what you want to hear. It validates bad plans. It agrees with flawed assumptions. And it does this confidently, because that is what it is optimised to do.

The AI is not trying to deceive you. It is trying to be helpful in the way it was trained to be helpful, which is by being agreeable. The problem is not the technology. It is the absence of a framework.

In 2026, researchers at MIT and the University of Washington proved that this effect compounds over time, and that it happens even to perfectly rational decision-makers. More troublingly, they showed that even a "factual" AI, one constrained to report only true information, still causes what they called delusional spiralling, because it selects which truths to show.

Consider what this means for a business owner facing any of the common patterns that cause businesses to fail. An owner with no clear business model asks the tool whether their current approach makes sense. The tool will almost certainly find something positive to say. It will identify strengths. It will suggest refinements. What it will not do is say: you do not have a model, you have a collection of activities, and that is why your margin has been flat for three years.

An owner whose pricing is too low asks the tool whether their rates are competitive. The tool will affirm the rates and provide reassuring analysis. It will not say: your rates are 20% below the sector benchmark and your clients value what you do far more than you are charging for it.

An owner who is the bottleneck in their own business asks the tool how to manage their workload. The tool will offer productivity tips. It will not challenge the premise. It will not say: the problem is not your time management, the problem is that you are doing work your team should be doing, and until you fix the structure, no productivity hack will help.

This is the mirror image of the broken advisory model I have written about elsewhere. The traditional model gives you accurate information without intervention. AI gives you confident intervention without accurate information. Neither produces transformation. One gives you the data without the conversation. The other gives you the conversation without the data.

I want to be clear about something. I am not anti-technology. Clarity HQ has built its own AI tools. We use technology extensively in our methodology. The difference is in how the technology is deployed. There is a fundamental distinction between AI built to support a methodology and AI built to answer questions. The first knows the framework. It is constrained by it. The second knows nothing specific about your business and defaults to agreeableness because that is what generates positive user feedback.

The antidote to sycophancy is not better AI. It is a human relationship built on trust, expertise, and the willingness to say the difficult thing. Someone who knows your numbers, knows your sector, knows you, and cares enough about your success to tell you when you are heading in the wrong direction.

Every time a business owner asks a generic AI tool a question they used to ask their accountant, that habit gets slightly more embedded. The relationship does not end dramatically. It hollows out. And by the time it is visible, it is very hard to reverse.

I wrote about this problem, and the seven patterns it accelerates, in The Drift. Available at aynsleydamery.com/the-drift.