Dan Herbatschek Talks How AI Is Changing the Way Leaders Think

Dan Herbatschek has built much of his career in the space where advanced technology and organizational behavior join, to know that the most significant changes artificial intelligence is producing are not the ones showing up in productivity reports. The efficiency gains are real and well-documented. 

Tasks that once required hours now require minutes as information that once demanded extensive research is available in seconds. Workflows that depended on specialized expertise are being automated at a pace that is still accelerating, but beneath all of that, something less quantifiable is happening. The introduction of powerful AI tools into professional environments has changed what leaders can do and begun to change what they are expected to do while more subtly pointing to what they believe their own role to be. 

When a tool can generate a strategic analysis, draft a communication, or synthesize months of data into a coherent narrative, the leader’s function inside that process necessarily changes. The question is whether leaders are examining that change deliberately or simply absorbing it without reflection, allowing the technology to redefine their role before they have decided what that role should be.

The Delegation of Judgment

The most consequential shift artificial intelligence is producing in leadership is the gradual delegation of judgment. Leaders have always delegated execution, and effective ones do so as a matter of principle, reserving their own direct attention for the decisions that most require it. What is new is the degree to which AI tools are now capable of producing outputs that look like judgment without the underlying cognitive process that genuine judgment requires.

In many domains, under many conditions, AI produces remarkably good recommendations. The risk is that leaders begin to relate to those recommendations the way they relate to the output of a trusted advisor. Judgment that is outsourced without examination is ratification, and organizations led by people who are ratifying as opposed to judging are exposed in ways that may not become visible until a novel situation arrives that the model was not designed to handle.

“The danger isn’t that AI will make leaders lazy,” Herbatschek says. “The danger is that it will make them confident in ways they haven’t earned. There’s a meaningful difference between a conclusion you’ve thought through and a conclusion that was generated for you and reviewed. Leaders need to know which one they’re working with.”

What AI Does to the Quality of Questions

One of the less-examined effects of AI integration in leadership contexts is what it does to the questions leaders ask. In a pre-AI environment, the quality of a leader’s questions was constrained by the cost of getting answers. Research took time. Analysis required effort. The friction involved in pursuing a line of inquiry meant that leaders had to prioritize. While often frustrating, that friction carried a side effect worth preserving in that it forced a degree of deliberateness regarding which questions actually mattered.

AI dramatically lowers the cost of getting answers, which is, on balance, a significant benefit. It also changes the incentive structure around questioning itself. When answers are cheap and fast, the temptation is to ask more questions, not better ones. The result, in organizations that have not examined this dynamic, can be an abundance of information and a poverty of genuine inquiry. 

Leaders are surrounded by answers to questions they asked reflexively, while the harder, more important questions go unformulated because formulating them well requires exactly the kind of slow, deliberate thinking that the AI-assisted environment discourages.

“A model can only answer the question you give it,” Herbatschek notes. “It cannot tell you whether you’re asking the right question. That part is still entirely human. And it’s the part that matters most.”

Maintaining Cognitive Ownership

The challenge for leaders in an AI-integrated environment is ultimately to maintain what might be called cognitive ownership over the decisions and directions that most define their leadership. Cognitive ownership requires understanding the reasoning well enough to interrogate it, while identifying the assumptions embedded in it and being genuinely capable of arriving at a different conclusion if the evidence warrants one. 

A leader who cannot do those things with a particular class of decisions has effectively delegated those decisions, regardless of whose name appears on the final approval. Maintaining cognitive ownership in an AI-rich environment requires deliberate practice. It means occasionally working through problems without AI assistance and asking, before reaching for an AI tool, what one’s own initial assessment of the problem actually is. AI outputs must be treated as a starting point for thinking, never a substitute for it.

The Leadership Capacities That Matter More, Not Less

Artificial intelligence is clarifying which capacities are genuinely irreplaceable in human leadership, so the abilities AI replicates well involve processing existing information according to established patterns. The abilities it replicates poorly are the ones that matter most in conditions of genuine novelty such as sitting with ambiguity without resolving it prematurely, integrating ethical considerations that resist quantification, and making decisions whose consequences extend far beyond the variables any model was trained to optimize.

“The leaders who will navigate this well are the ones who use AI to do more thinking, not less,” Herbatschek says. “The ones who struggle are the ones who use it to avoid thinking altogether and don’t notice the difference until it matters.”

Leaders who genuinely understand the distinction are using this moment as a prompt for clarity about what leadership is actually for. The time AI saves on routine cognitive work is, potentially, time available for the deeper engagement that has always been the most valuable thing a leader offers. 

The potential is only realized by those who treat recovered time as an invitation to think more carefully. Too many absorb it straight back into the acceleration that AI was supposed to relieve in the first place.