S1:E18 - AI Leadership Means Knowing When Not to Automate

Todd Curzon

Welcome to the show. Daniel, I want to begin with a sentence that makes ambitious executives slightly uncomfortable: the hardest AI decision is very often not adoption... it's restraint. Not whether you can automate it, but whether you SHOULD.

Daniel Carter

That word restraint is doing a lot of work. Because most leaders hear AI and they hear speed, scale, efficiency. They don't hear, "maybe leave that alone." So when you say restraint, what exactly are they protecting?

Todd Curzon

They're protecting judgment. And that is where the discernment premium begins. The discernment premium is the advantage a leader gains when they know when not to automate, not to delegate to a model, not to accelerate a process simply because a tool has made acceleration available. Access is cheap. Judgment is not. And in executive life, the cost of a bad yes is often far greater than the cost of a delayed no.

Daniel Carter

The bad yes. That's the phrase that sticks. Because I've coached enough senior leaders to know that most regrettable decisions don't begin as reckless decisions. They begin as enthusiastic ones. Somebody says, "We can roll this out in two weeks," and everyone feels clever. Then six weeks later they're cleaning up confusion, mistrust, rework... all the invisible taxes.

Todd Curzon

Exactly. There is a kind of busy AI enthusiasm right now that looks impressive from a distance. Pilots everywhere. Tool stacks multiplying. Internal demos. Breathless language. But if you stand back and look carefully, much of it is motion without hierarchy. Leaders mistake activity for maturity. They confuse experimentation with operating discipline.

Daniel Carter

Motion without hierarchy -- that's good. So the team is doing ten AI-flavored things, but nobody has sorted which one actually deserves executive attention.

Todd Curzon

Yes, and the quieter habit is much more valuable. It is the leader who says, "This workflow is repetitive and low-stakes -- fine, automate portions of it. This decision affects trust, accountability, or reputation -- slow it down." That person may appear less innovative in the meeting. But in the long run, they are building something far more durable. They are building institutional judgment.

Daniel Carter

Let me push on that a bit. Because some listeners are gonna hear this and think, "Fine, but restraint can become an excuse for fear." I've seen leaders hide behind caution when what they really mean is, "I don't understand the tool well enough to use it."

Todd Curzon

Fair. Not every no is wise. Some no's are simply disguised insecurity. But discernment is not reluctance. Discernment is specific. It can tell you why this use case is appropriate and that one is not. It can articulate the trade-off. It can say, "The output may be fast, but the verification burden is too high," or, "The draft is useful, but the final call must remain human because accountability cannot be outsourced."

Daniel Carter

And that difference matters because teams can feel it. They know when a leader is anti-tool, and they know when a leader is pro-judgment. Those are VERY different signals.

Todd Curzon

Very different. One creates fear. The other creates standards. And standards, in this moment, are worth more than enthusiasm.

Daniel Carter

Let me give you the number that should sober people up. Only 1% of organizations self-report AI maturity. One percent. Not twenty. Not ten. One. So when executives talk as though the market has already sorted this out, no -- most teams are still improvising.

Todd Curzon

That 1% is the entire story, really. Because it means access has spread much faster than judgment has. Tools are now easy to obtain, easy to test, easy to talk about. Readiness is not. A license is not a capability. A prompt library is not an operating model. And a company-wide announcement does not magically create managerial discernment.

Daniel Carter

The phrase I keep coming back to is false readiness. Since the tools sit right there in the browser, people assume the organization is further along than it is. It's like buying a piano and calling the house musical. The instrument arrived. The capability did not.

Todd Curzon

That's exactly right. The presence of the tool creates psychological overconfidence. People feel modern because they have access. But decision quality remains uneven. One team uses AI for first drafts and gets genuine lift. Another uses it in a customer-facing process with almost no governance and introduces risk they barely understand. Same access. Entirely different quality of judgment.

Daniel Carter

And the companies moving fastest are not always leading best. I think that's the tension executives hate admitting out loud. Speed photographs beautifully. Governance does not. Nobody posts a triumphant screenshot saying, "Good news, we decided NOT to automate a sensitive decision because the accountability chain was muddy." But that might be the most mature move in the building.

Todd Curzon

It often is. Because leadership at altitude is not measured by how many things you touched with AI. It's measured by whether the organization can use AI without creating confusion about ownership, quality, or consequence. If only 1% report maturity, then the competitive advantage is plainly not mere access. It is the discernment premium -- the ability to convert tool abundance into better executive decisions.

Daniel Carter

So if I'm a VP listening to this, and my CEO says, "We need to move faster on AI," what should I hear underneath that? Because sometimes "move faster" really means "don't get left behind," and sometimes it means "please make us look current."

Todd Curzon

I would hear two questions beneath it. First: where does speed actually improve the business? Second: where does speed merely increase the volume of unreviewed decisions? Those are not the same thing. The mature leader distinguishes between acceleration that compounds value and acceleration that compounds error. And if you cannot explain that difference to your team, then the team will improvise its own rules... which is usually how messes begin.

Daniel Carter

The improvisation point matters. Because once everyone is making private judgment calls in their own corner, you've got fragmentation. Ten small AI policies, none of them written down, all of them different.

Todd Curzon

Yes -- and fragmented judgment is one of the hidden costs of easy access.

Todd Curzon

This connects, for me, to something I see constantly in VP transitions. You were promoted for being the best Director. The most reliable person in the room. The one who caught every flaw, reviewed every deck, rescued every deadline. Then you become a VP and, without meaning to, you keep running the Director's playbook. Your team routes everything through you. You review everything before it ships. You work later and later, and your impact somehow feels thinner. That is the Director's ghost.

Daniel Carter

And AI can become a very elegant disguise for that ghost, right? Because now instead of saying, "I need to personally touch everything," the executive says, "I'll just use AI to process more of what comes through me." Same bottleneck. Faster bottleneck.

Todd Curzon

Precisely. AI becomes another form of personal load-bearing. Not leadership, just throughput. The competence penalty shows up here in a subtle way: the better you were at the old job, the more tempted you are to absorb complexity yourself. AI can make that temptation feel efficient. But if every important decision still terminates in your head, you have not scaled judgment. You have merely accelerated dependency.

Daniel Carter

That line -- accelerated dependency -- I'm gonna keep that one. Because I've lived a version of it. Years ago, before all this current AI frenzy, I was leading a team through a messy strategic transition. And I remember thinking, very sincerely, "If I can just hold the context in my own head a little longer, I can protect the quality." Noble story. Terrible system. What I was actually doing was training the team to wait for my synthesis.

Todd Curzon

That's a painfully honest distinction.

Daniel Carter

It was painful to notice. Because I felt indispensable. And leaders love that feeling more than they admit. The problem is, indispensability scales poorly. If I'd had today's tools back then, I might have made it worse. I might have said, "Great, now I can summarize more inputs, answer faster, stay on top of everything." But the issue wasn't my processing speed. It was my refusal to distribute judgment.

Todd Curzon

That's beautifully put. The bottleneck is rarely lack of information. It is often the leader's identity. Being the safe pair of hands becomes part of the self-concept. And then AI arrives and offers a seductive promise: keep being the hero, just with more horsepower. But the VP role demands something else. It demands altitude. Guardrails. Decision rights. Clear escalation. The move is from being the smartest responder to being the architect of how decisions get made.

Daniel Carter

So let me try to explain it back. The Director's ghost says, "I'll catch it all." The mature VP says, "I'll design a system that catches what matters." Is that close?

Todd Curzon

Very close. I'd only sharpen one word: not just catches -- discerns. Because in the AI era, the premium goes to the leader who can teach an organization what deserves human scrutiny and what does not.

Daniel Carter

Okay, let's make this concrete. Where should leaders be especially cautious -- not abstractly, but in actual executive work?

Todd Curzon

Four places immediately. High-stakes people decisions. Sensitive governance. Messy cross-functional trade-offs. And moments requiring trust-building. If you are deciding who gets promoted, how a delicate performance issue is handled, or how a reorg message lands, you are no longer just processing information. You are shaping meaning, consequence, and relationship. That is not an area for casual delegation to a model.

Daniel Carter

High-stakes people decisions -- that's the first one that should make people sit up. Because "promotion" sounds administrative until you remember a promotion decision can change somebody's income, visibility, confidence, whole trajectory.

Todd Curzon

Exactly. And sensitive governance is similar. If your name is on the line, if the organization may have to defend how a decision was made, then the standard changes. Then there are messy cross-functional trade-offs. Finance wants one thing, product wants another, legal sees risk, operations sees delay. AI can help summarize inputs, certainly. But it cannot own the trade-off. Leadership is the act of taking responsibility for the choice.

Daniel Carter

And trust-building -- that's the one I think people underestimate most. A team can tell when a leader is using a tool to avoid the hard human part. They may not say it in the meeting, but they feel it. If the moment calls for presence, nuance, maybe even apology... a generated draft is not the same thing as earned trust.

Todd Curzon

Yes. So the practical answer is a decision matrix. I tend to look at four variables: speed, stakes, reversibility, and accountability. Speed: does faster help here, or merely create rushed output? Stakes: what happens if this is wrong? Reversibility: can we unwind the decision easily, or is the damage sticky? Accountability: whose name, precisely, is attached when this goes sideways?

Daniel Carter

Stakes and reversibility together are powerful. Because a low-stakes, reversible task -- say a rough draft, internal summary, early brainstorming -- that's one category. A high-stakes, hard-to-reverse call with your name on it is a completely different animal.

Todd Curzon

Completely. Mature leaders make those distinctions visible. They don't keep them private. They create guardrails their teams can actually follow. Not grand policy documents nobody reads, but usable rules: where AI is encouraged, where it is reviewed, where it is prohibited, and when escalation is mandatory. The goal is not to slow the team down. It is to reduce ambiguity so people do not have to guess at the edge of risk.

Daniel Carter

And "actually follow" is important there. I've seen governance documents that are so polished, so complete, so unusable. Forty pages of caution, zero operational clarity.

Todd Curzon

Yes -- governance that reads beautifully and fails functionally. The better test is simple: can a busy manager use the rule in under thirty seconds? If not, the guardrail is decorative. And decorative governance is one of the faster ways to produce false confidence.

Todd Curzon

This is precisely why we built the AI Ready Executive cohort under Altitude. It begins June 1, 2026, and it is designed for VPs and Directors who can feel the gap between where they were promoted from and where the role now demands they operate. Twelve weeks. Cohort-based, online. Live and async. Not a course you simply watch, but structured work that helps leaders close that judgment gap under real conditions.

Daniel Carter

And I think the phrase that matters there is structured work. Because the leaders who struggle with this usually do not need another inspirational keynote on the future of AI. They need frameworks they can carry into Tuesday morning. They need peers. They need pressure-tested thinking.

Todd Curzon

Precisely. The program covers the competence penalty, the Altitude Decision Matrix, AI governance without becoming the bottleneck, the Stakes Escalation Model, and leading teams through resistance. In other words, how to stop operating like the best person in your old role and start leading at altitude. And just as importantly, it gives you peer accountability -- other leaders navigating the same transition, which matters more than people realize. Executive judgment sharpens in conversation, not in isolation.

Daniel Carter

The peer piece is underrated. If you're a VP and everybody around you assumes you should already know this stuff, the loneliness gets expensive. You start improvising in private. And given that only 1% of organizations self-report AI maturity, private improvisation is... probably not the best development plan.

Todd Curzon

No, it is not. Which is why we also have the free live session on The Competence Penalty on April 29 and April 30, 2026 -- same content, two times -- for leaders who want to understand the pattern first. Because I do think people deserve a sober diagnosis before they commit to a solution. The point is not to sell urgency. The point is to help capable leaders see, clearly, where the Director's ghost is still running their day.

Daniel Carter

And that, to me, is the credible invitation here. Not "come learn the hottest tools." More like, "come learn how not to become the bottleneck in an era that rewards speed and punishes sloppy judgment."

Todd Curzon

Yes. Because in the AI era, the premium does not belong to the loudest adopter or the busiest experimenter. It belongs to the leader who can tell the difference between useful acceleration and expensive overreach.

Daniel Carter

That's the work. We'll leave it there.