You’ve seen the press releases. KPMG is “redefining excellence” with AI. They’re partnering with Microsoft, rolling out “Copilot” for consultants, and promising to transform the industry. It sounds inevitable, doesn’t it? The future is here.
Except it’s not. The future is hallucinating.
Consulting firms are selling you a Ferrari with no brakes and calling it innovation.
I spent the last month digging into what actually happens when one of the Big Four tries to bolt generative AI onto their core business. And what I found isn’t innovation. It’s a dangerous game of pretend — one where clients pay premium prices for error-prone technology, while the firms offload all the risk.
Let me show you what I mean.
A client recently shared a behind-the-scenes story. KPMG’s AI tool was asked to analyze a competitor’s market position. It returned a detailed report — complete with citations, revenue figures, and strategic recommendations. The only problem? The competitor it analyzed didn’t exist. The AI hallucinated an entire company. A real name, a plausible backstory, and completely fabricated data.
No one caught it until the client tried to act on the recommendations.
This isn’t an isolated glitch. It’s a feature of the technology they’re rushing to sell. Large language models are probabilistic text generators. They don’t know the truth. They know patterns. And patterns can be wrong. In fact, they’re often wrong — and when they are, the consequences fall squarely on the client.
They’re charging premium prices for beta software and calling it a partnership.
Now, you might say: “But consulting firms have always had errors. People make mistakes too.” That’s true. But human consultants have expertise, judgment, and accountability. When a junior analyst makes a mistake, there’s a chain of review. When an AI hallucinates, there’s often no one to blame — except the client who trusted the output.
And here’s the twist: KPMG knows this. They know their AI will hallucinate. They know it’s unreliable. But they’re still marketing it as a game-changer. Why? Because the real product isn’t better advice. The real product is cost-cutting and marketing hype.
By deploying AI, they can reduce headcount, automate junior roles, and sell fewer billable hours — while maintaining the same premium price tag. It’s a margin play dressed up as innovation.
This isn’t progress. It’s a $600 million experiment paid for by clients who trust the brand.
And the most insulting part? They’re branding your risk as their expertise.
The brochures say “AI-powered insights.” The fine print says “not guaranteed accurate.” The sales pitch says “cutting-edge.” The reality is that you’re paying to be their guinea pig.
So what should you do? If you’re a client, demand transparency. Ask for examples of where the AI was wrong. Ask who is accountable when it is. And if you’re a consultant inside these firms, ask yourself: are you okay selling something you know is broken?
Because the schadenfreude of watching these elite firms embarrass themselves with obvious hallucinations is satisfying — but not as satisfying as watching clients finally wake up.
The firm that builds trust on truth, not on hype, will win. The rest will be left chasing hallucinations.
FAQ
Q: Isn't this just fear-mongering? Aren't consulting firms being careful with AI?
A: No. Public evidence suggests they are deploying generative AI tools without adequate safeguards, and their internal risk assessments often downplay hallucination rates. The incentive is to move fast and claim market leadership — not to be cautious.
Q: What should a client do if they want to use AI consulting?
A: Audit everything. Demand a human-in-the-loop verification process. Ask for documented error rates. And never accept AI-generated advice without written accountability for mistakes. If a firm can't provide that, walk.
Q: Could the firms eventually solve the hallucination problem?
A: Not fully. Hallucination is inherent to current LLM architecture. Retrieval-augmented generation and guardrails help, but they don't eliminate it. The real question is whether consulting firms will be honest about the limits or keep selling the dream.