The End of Ignorance Tax

The Knowledge Barrier is Gone

A few years ago I was handed a role I was not ready for.

I’d had success before that. Moved fast, read rooms, made calls. That version of the job rewarded initiative. You saw the problem, you solved it, you moved on. I was good at it.

This new role was different. It required structure. Methodology. Deliberate, process-driven leadership that doesn’t come naturally when your instinct is to just handle the thing in front of you. And because it was senior, there was nobody above me to pull me aside and explain how it actually worked.

So I kept moving. That’s all I knew how to do.

Here’s the thing nobody tells you: not knowing the answers is manageable. Not knowing the questions is a different problem entirely.

You can search almost anything. But you can’t search what you don’t know you’re missing. I didn’t have the vocabulary yet. I didn’t have the mental model for how these decisions were supposed to be made. I was building the plane while flying it with no co-pilot and no manual.

Mistakes followed. Some were painful. All of them were expensive in time, in relationships, in credibility. The lessons came eventually, but they came after the damage was done.

That era is over.

If you dropped me into that same role today, or any role I’ve never held, I’m not worried. Not because I’m overconfident. But because the gap that made those early years so hard no longer exists in the same way.

The models we use right now are not fancy autocomplete. GPT-4o now scores above 90% on each step of the USMLE, the medical licensing exam that filters who gets to practice medicine in the United States. Microsoft’s AI diagnostic system recently outperformed experienced physicians on complex medical cases by a factor of four. On PhD-level science benchmarks, leading models now exceed average human expert performance. OpenAI ran a real-world productivity study evaluating AI output against professionals with an average of 14 years of experience across 44 different occupations. The best models matched or outperformed those experts nearly half the time.

That’s not the ceiling. That’s where things stand right now.

In practice, I have a subject matter expert in almost any domain available at any hour who can actually talk back. Not a search result I have to interpret. Not a consultant I have to schedule. Something that meets me where I am.

But the part that matters most isn’t just getting answers. It’s getting the questions right.

One of the most useful things I do now is tell AI to reverse-prompt me. I lay out the situation and say: ask me the questions I’m not thinking to ask.

It does. The questions a lawyer would ask before signing. The questions a CFO would ask before committing. The questions a veteran operations leader would ask before restructuring a team. Blind spots I wouldn’t have known to look for, surfaced in two minutes.

That’s the thing that would have changed everything for me earlier. Not just having answers available. Having a way to discover what I didn’t know.

When I was building a division from the ground up there were stretches where I was operating well outside my experience. Pricing models I’d never built. Contracts I’d never structured. Org design for a team that didn’t exist yet. I figured it out, but I figured it out the hard way.

More recently I’ve built AI systems for a federal agency including a full compliance system presented to senior leadership. That work required navigating compliance frameworks, procurement constraints, and technical architecture decisions all at once. As a non-technical resource, these were domains that didn’t overlap cleanly with anything I’d done before. I used AI as a thinking partner throughout: give me the framework, show me the failure modes, ask me what I haven’t considered.

And in some cases AI doesn’t just advise. It builds. Drafts contracts. Structures proposals. Writes code. Things that used to require hiring a specialist or spending weeks getting up to speed.

The result is that I say yes to more things now.

A few years ago I’d hesitate before taking on something outside my experience. Not because I was afraid of the work. Because I knew how painful the ignorance tax was. The cost of not knowing what you don’t know.

That cost is lower now. Dramatically lower.

When someone asks me to step into something unfamiliar, my first move isn’t to think about what I don’t know. It’s to think about what I can learn in the next hour. That’s usually enough to get started. And getting started is most of the battle.

The going from zero to one used to be the hardest part. Now it’s just the first conversation....