Where AI Ends and You Begin: Why Human Judgment Is About to Get More Valuable

TL;DR: The "replace humans with AI" era is already unwinding — faster than most predicted. Companies that over-cut are quietly rehiring, because they gained speed but lost the signal. And the data backs it: human judgment is getting more valuable, not less. This is about where AI ends, where you begin, and why that line is the biggest opportunity of the decade — for anyone willing to stop being afraid and lean into what makes them human.

Not long ago, the plan at a lot of companies was simple: replace people with AI. That experiment is already failing — and honestly, faster than I expected. Companies that made deep, AI-driven cuts are quietly rehiring, because they've run into a hard truth: technology enhances human dynamics. It doesn't replace them. They picked up the speed, and they lost the signal.

So the real question was never "will AI replace me?" The real question is: where does AI end, and where do you begin? That's the line worth understanding — because getting it right is the whole opportunity.

The 20-watt machine you're underestimating

Here's something worth sitting with. The human brain runs on about 20 watts. That's almost nothing. A person can function — maybe not optimally, but function — on Diet Coke and Taco Bell. Low energy in, staggeringly high-quality output out. Now compare that to the sheer compute and energy it takes to run the systems we're told are going to replace us. It isn't close.

The human brain is extraordinarily dynamic, and I think we've lost sight of what it's actually for. We romanticized the post-industrial era — we got comfortable with work that turned people into cogs — and in the process we forgot the true potency of what our minds can do. In that narrow sense, people who fear AI are right to feel uneasy. But they're worried about the wrong thing.

I had a conversation recently with someone who was scared: a company had pitched that AI could do exactly what she does — essentially a glorified bean-counting role — and she looked around and said, "Well, what am I going to do?" And here's what struck me. The fact that we've accepted that those are jobs we even want means we've lost our sense of pushing cognitive boundaries. We've dulled our appetite for exploration, critical thinking, judgment, wisdom, cross-domain synthesis — taking an idea from one place and applying it somewhere it doesn't obviously belong. Watching a movie and getting inspired. Having a conversation that hands you a light-bulb moment. That's what our brains were built for. That's the work that actually fulfills us.

Why I've never worked more — and I love it

Here's the part people don't expect me to say: I've never worked more hours than I do now, post-AI. And I love it. Not because I'm grinding — because AI removed all the friction that used to exhaust me. It cleared the low-value drudgery off my plate and freed me to spend my energy on the work only I can do.

We're in a rediscovery period. And it's normal to cling to the past — the past feels safe, the known feels safe. Everybody's had their existential crisis about this; I lost plenty of sleep and spent months investing in understanding it before I'd speak on it with any confidence. I don't put things out I don't stand behind just to earn a click or a lead — my integrity won't allow it. So let me show you what the numbers actually say, because they tell a very different story than the fear does.

The data: humans are getting more valuable, not less

Start with my own field. The coaching profession went from about $2.85 billion in 2023 to roughly $5.34 billion in 2025 (ICF/PwC data) — nearly doubling, in the exact window that ChatGPT went mainstream. If AI were the replacement, that number should be cratering. Instead it exploded.

Now the stat that matters most. According to PwC's Global AI Jobs Barometer, which analyzed over a billion job ads, judgment-heavy roles are growing twice as fast and paying 42% higher wage growth than the automatable ones. Read that again. The market is pricing judgment up. So the person who just wants the bean-counting job is missing the real opportunity — because while that task is being automated, the work that pays 2x faster growth is sitting right there, waiting for someone to step into it. Especially if you use AI to enhance the work you're actually great at.

The trust gap AI can't close on its own

Here's why humans keep winning: AI has a trust problem and an accuracy problem, and neither is going away soon.

Only about 10% of people trust the first answer they get from AI, and roughly 48% verify it against multiple sources (Yext). Almost half of people are going somewhere else to double-check. That happened in my own house this week — my wife ran a simple credit-card question through three separate AI threads and got three different answers. One was right. Same question, same clean data, three answers.

And it goes deeper: AI search fails to retrieve correct citation information more than 60% of the time, across roughly 1,600 tested queries (SearchSignal). Part of why? AI is increasingly referencing AI-generated content to answer you. In other words, it's getting high on its own supply.

None of this is me poo-pooing AI — it's the most powerful tool I've ever used, and I've got a Claude thread open nearly every waking hour (I'm using one to produce this very piece). The point is simpler: when the stakes are real, people don't fully trust the machine. They come back looking for a human they know, like, and trust.

The paradox has a name: the Human-AI Handshake

So here's the paradox. People are paying more for coaching, committing for longer, and in some cases flying across the world to work with someone in person. Why would they do that if AI already gave them everything they needed?

The answer is what people are starting to call the Human-AI Handshake: you pair the speed, scale, and processing power of the machine with the things only humans bring — intuition, judgment, discernment, wisdom, lived experience, and real domain expertise. You catalyze all of it by using the tool to create velocity. That's the whole model. And it's exactly why companies are running back to hiring: they realized they'd picked up the speed but lost the signal — and that was a critical mistake.

Someone I deeply respect on this is Jensen Huang, the CEO of Nvidia, who called it early. His view, stated repeatedly and publicly: you won't lose your job to AI — you'll lose it to "somebody who uses AI." And notably, he's said that the more AI his own company uses, the more people they end up having to hire — because these systems create entirely new skills and expand what's possible. The companies that win aren't the ones replacing humans with AI. They're the ones pairing humans with AI. Put those two against each other, and it isn't a fair fight.

Proof: I built the jet — but I still had to fly it

I don't like making claims I can't demonstrate, so here's what this looks like in practice.

I recently migrated my own website — a cluttered WordPress site with broken links and messaging that no longer matched my brand. I exported the files, had Claude analyze them, and rebuilt it as a custom React site in a single Sunday afternoon. It's the best and highest-performing site I've ever had, measured by Google's own page-speed scores — and I had it audited independently by my CTO and by a fresh Claude thread built specifically to run audits. Both said the same thing: this is a professionally done migration. I also built a custom app for our community in about 15 days — something engineer friends told me would take a full development team six months.

I could not have done any of that before AI. But here's the part that matters: I had to project-manage the entire thing. It wasn't "Claude, go build me a website." It wasn't "Claude, go build me an app." That's the expectation we're moving away from — the one that had companies laying people off assuming the machine would just do it, only to discover they needed pilots.

Because that's what we built: incredible, F-18-caliber fighter jets, digitally speaking. And it turns out they don't fly like they should without a pilot. We've commoditized intelligence. We have not commoditized the dynamic aspects of what the human mind can do — and I don't believe we ever will.

Even our dystopias know this

Here's a thing I can't unsee. Even in the darkest sci-fi — i, Robot and the rest — the humans are still the pilots. Still the leaders. Even the police force is led by people. Why? Because even in our wildest dystopian imaginations, there's something we innately understand: humans belong in the seat of judgment, discernment, wisdom, synthesis, and reading context — the verbal and the non-verbal. Those aren't things you program.

I'll be transparent about my own lens here, because it shapes how I see this. As a believer, I think the dynamic human mind is ordained, not engineered. We created AI — and everything we've ever built to try to match what was already created has fallen miserably short. Consider: if we could harness what a single tree pulls from the sun, we could power a city — and yet we still live with energy deficits. We've come up short on far simpler things. So the idea that we've suddenly built something that supersedes human ability doesn't hold up for me. The machine is a remarkable partner. It is not a replacement for the thing that makes you you.

What to actually do about it

Quit worrying. Quit being afraid. Start adopting, adapting, embracing, and building — and run toward the opportunity instead of bracing against it.

Practically: hand AI the friction — the execution, the repeatable tasks, the drudgery that used to exhaust you. Bring your judgment to the work that actually requires a human. Use the speed to do more of what only you can do. And understand that you're not behind for still valuing human judgment — you're early to the thing that's about to get more valuable, not less.

Lean into what makes you human. Not the cog-work we were forced into that took us away from being human in the first place. That's where AI ends. That's where you begin.