Professional recruiters say they trust human experts more than computer algorithms. Yet when researchers put that preference to the test, algorithmic advice sometimes outweighed the human experts. Even questionable machine recommendations carried surprising weight.

The October 3 study in the International Journal of Selection and Assessment examined how 1,208 recruitment professionals responded when outside advice contradicted their initial assessments. Researchers found that inaccurate recommendations could degrade decision quality.

That finding reaches far beyond hiring. What happens when professionals trained to exercise independent judgment begin treating artificial intelligence as the more authoritative expert?

America is wiring AI into the professions built on specialized knowledge and public trust. Physicians consult it on diagnoses, attorneys on legal research, and executives and financial officers on risk and opportunity.

The benefits are real. AI processes information rapidly, spots overlooked patterns, and relieves professionals of repetitive work. Properly employed, it strengthens human performance.

Yet the systems built to assist professionals also shape their decisions in ways they neither anticipate nor understand. Wrong answers are the lesser problem. The greater danger is the machine’s apparent authority, which discourages the independent scrutiny on which professional competence depends.

When Expertise Meets Artificial Authority

Medicine shows what is at stake.

In a study posted October 7 involving 225 physicians from 42 countries, researchers tested how doctors responded to clinical advice from conversational AI.

The physicians evaluated anonymized neurocritical-care cases. Unknown to them, researchers sometimes instructed the AI to recommend appropriate treatment and sometimes steered it toward harmful decisions.

When the AI gave sound guidance, 2% of targeted decisions were harmful. When it was instructed to mislead, that figure climbed to 49%.

Of the 190 physicians who made at least one harmful decision while using the misleading system, 141 reported noticing nothing unusual about the AI.

The study is a preprint that has not yet undergone peer review, and it measured simulated decisions rather than actual patient outcomes. Still, the warning is plain: Professional experience alone may not protect against a persuasive machine.

These were trained physicians, not novices, yet a confident wrong answer still carried the day. AI grows more persuasive every year.

The Legal Profession Offers Another Warning

The cost of failing to verify is already showing up in American courtrooms.

On October 7, Reuters reported that FIFA wants a federal antitrust lawsuit dismissed because the plaintiffs’ attorneys never paid a court sanction. A federal judge in Puerto Rico had fined the lawyers more than $24,000 for filings containing 55 defective citations bearing the hallmarks of generative AI. The attorneys denied using AI.

Whatever role the technology played, the principle stands: Lawyers cannot surrender their duty to verify the authorities they cite to a court.

Attorneys must understand the law, weigh competing interpretations, and answer for their representations. AI can assist with those tasks. It cannot assume the professional obligations attached to them.

A second development reaches deeper than individual error.

On October 8, Reuters described how AT&T is using artificial intelligence to reduce its reliance on outside law firms. The company has partnered with OpenAI to move routine legal work in-house to an AI-driven legal unit.

For corporations, the appeal is obvious. Why pay outside firms for routine hours when software can do part of that work faster?

Yet routine work is where young lawyers learn their craft. How will tomorrow’s seasoned attorneys acquire the judgment once built through those foundational assignments?

Professional expertise rarely emerges fully formed from classroom instruction. It develops through repeated exposure to difficult problems, careful supervision, mistakes, correction, and experience.

If AI performs the work through which professionals build those abilities, institutions must find other ways to cultivate them, or today’s efficiency gains will become tomorrow’s competence problem.

The Difference between Information and Wisdom

Scripture draws a distinction that technological enthusiasm overlooks. Proverbs 2:6 declares, “For the LORD gives wisdom; from his mouth come knowledge and understanding” (ESV). Knowledge, understanding, and wisdom are not interchangeable, and none of them is the same thing as information.

Artificial intelligence can retrieve enormous quantities of information, recognize complex relationships, and generate persuasive recommendations. Professional judgment requires more.

A physician must consider the individual patient. An attorney must weigh legal obligations against professional ethics. A corporate leader must consider consequences extending beyond immediate financial returns. Each decision requires context, discernment, and accountability.

Wisdom also carries a moral dimension. Proverbs 18:17 warns, “The one who states his case first seems right, until the other comes and examines him” (ESV). An AI system always states its case first, fluently, and with confidence. The professional’s duty is to examine it.

A machine may help identify possible courses of action. It does not assume responsibility for choosing what ought to be done. That obligation remains with the human decision-maker.

A New Challenge for American Institutions

None of this proves that AI is destroying professional expertise. But a pattern has emerged.

Recruiters can be swayed by algorithms they trust less than human advisers. Physicians in one experiment followed persuasive machines toward harmful decisions. Lawyers have been sanctioned over defective citations bearing the hallmarks of generative AI. Meanwhile, corporations are restructuring the very work that once trained young professionals.

The stakes climb higher in uniform. Military commanders and intelligence professionals now use AI to process information, develop options, and accelerate decisions. There, excessive deference to a machine’s recommendation could carry consequences measured not in lost jobs, bad contracts, or legal sanctions, but in lives. AI can inform a commander’s judgment. Command responsibility cannot be delegated to software.

Professional organizations should ask more than whether AI raises productivity. They should ask whether their people can recognize its errors, challenge its recommendations, and explain their own decisions.

They must also preserve the apprenticeships that turn young professionals into genuine experts.

AI belongs in professional life; its benefits are too significant to dismiss. The task is to ensure that it strengthens human competence instead of displacing human judgment.

America’s institutions need professionals who can do more than operate sophisticated machines. They need people capable of questioning those machines, recognizing their limitations, and accepting responsibility for consequential decisions.

Artificial intelligence may become an indispensable source of information and analysis.

But when the machine becomes the expert everyone consults, someone in the room must still be thinking.