Companies have a new problem when deciding who is ready to lead. Imagine two directors competing for the same promotion. One has a strong understanding of the business, spots problems early and consistently makes sound decisions. The other is capable but less effective at analyzing complex situations and identifying the right course of action. Yet both arrive at the promotion meeting with thoughtful recommendations, all refined with AI. Their track records still matter, but when their work looks remarkably similar, those recommendations may reveal less about which director is better prepared for the next role.

This creates what we might call visible competence compression: AI can narrow the observable difference between average and exceptional performance without necessarily closing the gap in actual ability. As the quality of AI-assisted work improves, companies may find it increasingly difficult to distinguish polished output from genuine leadership capability. That raises questions about how organizations evaluate talent, identify future leaders, and decide whom to promote.

A 2026 study published in Management Science found that generative AI improved the quality of pitches in hiring and startup-investing experiments while making it harder for employers and investors to distinguish genuine expertise. Evaluators’ screening accuracy increased by an estimated 4% to 9% when participants had access to AI. The study did not examine corporate leaders, but its findings raise questions about how companies evaluate leadership potential.

AI May Improve The Work Faster Than It Improves The Leader

None of this means AI-assisted competence is somehow fake. AI can make people genuinely better at their jobs, including experienced professionals.

In research published in Organization Science in 2026, 791 professionals at Procter & Gamble worked on real product-development problems. Individuals using AI produced work comparable in quality to two-person teams working without AI. AI also helped employees move beyond their usual functional specialties, with technical and commercial professionals producing more balanced solutions when they used the technology.

AI can also help capable people whose writing or presentation skills previously obscured their expertise. In the Management Science experiments, it improved evaluators’ ability to recognize expertise in some settings involving candidates from non-English-speaking countries. The challenge is distinguishing when AI reveals talent that was already there and when it makes limited capabilities harder to recognize.

A September 2026 National Bureau of Economic Research working paper examined this difference. It reports results from a three-month randomized trial involving 133 patent lawyers at 11 U.S. intellectual property firms. Lawyers given an AI drafting assistant produced higher-quality work after both 10 and 90 days, with larger immediate improvements among junior lawyers.

After three months, researchers tested the lawyers’ professional judgment without AI. Senior lawyers who had used the technology outperformed the senior control group, but junior lawyers showed no average improvement in their unaided judgment. The authors concluded that foundational expertise may matter for turning AI assistance into durable professional skill. Six of the seven researchers were affiliated with Google, which developed the tool used in the experiment.

Patent lawyers are not executives, but the study raises a relevant question: Is AI developing managers’ analytical abilities or primarily improving their output? Both benefit the organization, and using AI effectively is a valuable skill in itself. Neither automatically establishes readiness for greater leadership responsibility.

When Leadership Promotions Reward Polish Over Capability

Let’s return to the scenario of the two directors. Both can use AI to interrogate a strategy and prepare talking points before an executive meeting. How does the promotion committee know which person has the stronger judgment?

The easy answer is outcomes. Promote the person who delivers better results. Real organizations are rarely that straightforward. Results are influenced by the quality of someone’s team, budget, market conditions, timing, inherited problems and luck. Promotions also happen before companies have years of clear evidence about how someone will perform at the next level.

The problem is that companies have spent years rewarding something loosely called executive presence. The phrase can encompass legitimate abilities, including confidence, composure and communication. But those qualities can also make it easy to confuse someone who presents an idea convincingly with someone who has chosen the right idea in the first place.

AI makes that confusion harder to ignore. An uneven writer can produce elegant prose. A rambling thinker can ask a model to impose structure on an argument. The finished work can look impressive even when the reasoning behind it has not improved to the same degree.

AI did not create the flaws in how companies evaluate leaders. It may simply be exposing how often they have mistaken polish for competence and confidence for the ability to lead. Communication remains an important leadership skill, but presenting a strategy convincingly and choosing the right strategy deserve separate consideration.

When traditional indicators become less revealing, companies may fall back on other ways of deciding who looks ready to lead.

Columbia Business School researcher Nataliya Wright has warned that when useful evidence of expertise becomes harder to interpret, decision-makers could rely more heavily on other characteristics, including educational pedigree or demographic attributes.

And the consequences extend beyond the two candidates. A promotion often gives someone greater influence over budgets, employees and strategic decisions. If a company mistakes AI-assisted polish for the ability to handle those responsibilities, the difference may only become apparent once the person is in charge.

This is where compression becomes more than a talent-identification problem. It becomes a question of whether companies are evaluating the capabilities required for the next role or simply rewarding how convincingly someone performs in their current one.

How Companies Can Evaluate Leaders After Compression

As polished answers become easier to create, companies will need to pay closer attention to how leaders arrive at decisions. Can they identify the problem worth solving? Explain why they rejected a plausible alternative? Recognize when an impressive analysis is addressing the wrong business problem?

Leaders have always relied on analysts, chiefs of staff and other specialists. AI is another form of assistance, and using it effectively should not count against someone. The better evaluation is whether the person understands and owns the thinking behind the finished work.

  • Ask which alternatives they considered, why they dismissed them and what evidence could change their decision.
  • Present a realistic business problem, introduce new information and observe how the leader reassesses their recommendation. The goal is not to reward the quickest answer. It is to understand how the person thinks, incorporates evidence and decides when a different course of action is warranted.
  • Pay attention to whether the leader repeatedly identifies important problems early, makes useful tradeoffs and adjusts when evidence changes.
  • Examine whether the leader translates decisions into action, helps the team address obstacles and takes responsibility for the outcome.

These practices should supplement, not replace, evidence of actual performance. A promotion decision still needs to account for business results and the responsibilities of the prospective role.

The technology may ultimately improve leadership evaluation by forcing companies to reconsider what they have been rewarding. AI may raise the quality of work average leaders can produce, while helping excellent leaders become even better. But as the visible differences narrow, companies will need to look beyond the finished product to determine who can lead people and handle greater responsibility.