Who checks AI’s facts?

A quiet storm is gathering in the world of corporate advisory. Recently, a major report commissioned by a national government was found to contain numerous factual errors, including references to documents and studies that did not exist.

What made the case unusual was not just the scale of the mistake but the method behind it. Much of the report had been drafted using artificial intelligence (AI), and the errors went unnoticed until after publication.

The episode raised uncomfortable questions about whether consulting firms, long trusted as guardians of accuracy and insight, are now too dependent on the very technology that could one day replace them.

AI is designed to assist, to summarize, to make sense of vast quantities of data. It is fast, efficient, and persuasive. But its confidence often masks its fragility.

When AI systems are asked to generate text, they do not understand truth in the way humans do. They assemble words that seem plausible, borrowing patterns from what they have been trained on.

Without human oversight, the result can be sentences that sound factual but have no grounding in reality. In the case that captured public attention, the AI produced convincing references that turned out to be fabricated.

The consulting team relied on those references without checking them, and in doing so exposed a flaw that goes beyond software. It showed what happens when convenience replaces diligence.

The incident is not just a cautionary tale about technology. It also challenges the economic and ethical foundations of the consulting industry itself.

For decades, large firms have sold expertise and judgment, promising clients that every conclusion they deliver is backed by rigorous research. But what happens when those same organizations quietly depend on generative AI to speed up their work?

If a polished report can now be written by a machine, what exactly are clients paying for? Some may argue that the value lies in interpretation and experience, that human consultants still provide context that algorithms cannot.

Yet this argument loses weight when human experts fail to verify even the basic facts that underpin their recommendations.

There is also a deeper risk that goes beyond embarrassment. Many of these reports influence public policy, spending decisions, and national programs. A single error can distort a strategy, misinform a reform effort, or mislead decision makers who trust the consultant’s authority.

The responsibility to validate information, therefore, is not optional. It is the price of credibility. In a time when AI can produce convincing but false data at scale, the burden of truth checking must grow heavier, not lighter.

The larger question, then, is not whether AI will replace consulting firms, but whether consulting firms can adapt to coexist responsibly with AI. Technology is not the enemy.

Used correctly, it can improve efficiency, surface insights, and even democratize access to analysis. The danger lies in using it as a shortcut rather than a support tool. Organizations that fail to build stronger quality controls, audit mechanisms, and disclosure policies risk eroding the very trust that sustains their business.

Artificial intelligence has transformed the way we write, research, and reason, but it has not replaced the timeless rule that truth demands human care. When people stop questioning what machines produce, they surrender the very craft of thinking.

The recent controversy is a warning that in an age of smart tools, real intelligence is not about speed or polish but about doubt, discipline, and verification. In the end, credibility cannot be coded. It is earned, line by line, by those who still choose to ask if what was written is indeed true.