Rejected in Minutes: Why AI Cannot Judge a Leader
Over the past few months, I have spoken with a remarkable number of highly experienced executives who are looking for their next role. Former CEOs, CFOs, CHROs, managing directors. People with 20 or 25 years of leadership behind them, turnarounds, international expansions, board experience.
Many of them tell me the same story. They carefully choose a handful of roles that are a good fit with their profile. They send their CV. And within minutes, sometimes in the middle of the night, an automated rejection arrives. No human has read a single line.
This cannot be right.
What the machine actually sees
Applicant tracking systems or broader HRIS systems like Greenhouse, Workday or SAP Success Factors and others are very good at one thing: extracting data points - job titles, company names, dates, keywords. They compare these against a set of criteria and produce a ranking.
What they cannot see is everything that actually matters at senior level. How did this person handle a crisis? Where are their real strengths, and where are the blind spots? What is their potential for the next, bigger role? A system that reads a CV like a spreadsheet has no judgment. But judgment is precisely what is needed when we appoint a leader.
The problem gets worse the more senior someone is, because senior careers are rarely linear. Board mandates alongside an executive role. An interim assignment. A sabbatical. A portfolio phase. A step sideways that was, in hindsight, strategically smart. The system reads a gap, or "overqualified". A human reads a story and puts data in perspective.
Even the hard facts need context. "Grew revenue by 30%" means something very different in a booming market than in a declining one. An achievement only becomes meaningful against the situation someone inherited, and no parser can reconstruct that.
Meanwhile, we are drifting into an absurd arms race. Candidates increasingly use AI to tailor their CVs to beat the filter, while companies use AI to screen them. Two machines negotiating over keywords. When both sides optimise for the algorithm, the CV stops telling you anything about the person. Most executives I speak with refuse to play that game, and frankly, they shouldn't have to.
The legal question is already on the table
There is also an uncomfortable dimension here. Experienced executives are, almost by definition, older. In the US, the case Mobley v. Workday is testing exactly this. In February 2026, a federal court in Northern California authorised notice to potential members of a collective action alleging that AI-driven hiring software may have unlawfully screened out applicants aged 40 and older. The court also rejected Workday's argument that these protections only apply to current employees, confirming that applicants over 40 can challenge AI screening tools. There is no finding of liability yet, but the question will not go away.
In Europe, the AI Act already classifies AI used in recruitment as high-risk, with requirements such as human oversight. Following the Digital Omnibus, those obligations for recruitment systems have been postponed from August 2026 to 2 December 2027. Deferred, not cancelled. Companies would be wise to use the time rather than wait for it.
And in executive search?
I want to be clear: I am not against AI. AI is changing our industry profoundly. In our own work, it has become highly useful, especially at the beginning of a search. It is speeding up and supporting desk research, market mapping, identifying relevant companies, matching potential candidates against defined criteria. Tasks that used to take a researcher weeks can now be done in hours or days, and AI can help stress-test assumptions. That is real progress, and it benefits boutique firms in particular, because it levels the playing field in research against the large global players.
But three things remain true.
First, the output needs a human. AI produces lists, not answers. Someone with market knowledge, judgment and plain common sense has to closely review every name, remove the obvious mismatches, add the people the machine missed, think outside of the box, draw conclusions backed by market knowledge. Very often, the most interesting candidate is not the one with the perfect keyword profile. And this still takes time – even if the AI has taken out a chunk of time.
Second, the results are hard to reproduce. Run the same search twice with the same criteria and you will often get a different longlist. For a process that should be rigorous and defensible towards a client or a supervisory board, that is a real limitation. These tools are assistants, not authorities.
Third, and most importantly, the decisive part of the work begins where AI ends. The conversation “Who is actually sitting across from me? How does this person think, listen, react under pressure? Is there credibility, empathy, authenticity? Does the energy match the situation the company is in? What is my gut feeling telling me, and is it backed by what I hear?” Recognising strengths, weaknesses and potential in a person is not a data problem. It is a human one.
The decisive part of the work begins where AI ends.
And a search runs in both directions. The best candidates are usually not looking. Someone has to explain the opportunity, address their concerns, and read whether they are really interested or just curious. AI can find a name. It cannot win a candidate.
The same applies to reputation. The most valuable insights about a leader rarely come from a database. They come from informal conversations: how someone treats their team, how they behaved when things went wrong, what former colleagues say when they speak openly. That knowledge lives in networks, not in data.
What this means for companies
If your organisation uses automated screening for senior roles, ask yourself a simple question: How many excellent leaders have you rejected without anyone ever reading their CV? Efficiency at the entry point might be valuable. At the top, it can be misleading and very expensive.
Efficiency at the entry point might be valuable. At the top, it can be misleading and very expensive.
AI should make us faster and better informed. It should not make the decision. For leadership roles, the final judgment belongs with people who know what good leadership looks like, and who take the time to sit down and find out.
AI should make us faster and better informed. It should not make the decision.
I am convinced that while assisting us in the process, AI will not replace the advisor and the human element, and that it cannot discover talent, especially where there is little to find online.
And for executives in transition
I believe portals are the least effective route to your next role. Invest in your network, in direct conversations with board members and shareholders, and in relationships with search consultants who actually read your profile.
At this level, the right role almost always comes through a person, not a portal.
ABOUT THE AUTHOR
Marion Heil is founder and managing partner of Board+CEO Advisors, a Vienna-based high-end executive search and board advisory boutique. She advises listed companies, family businesses and investors on C-suite, leaders and supervisory board appointments and succession mandates across DACH and EMEA.
