When the AI Skeptic and I Agree, You Should Worry

I just found myself nodding along to Cal Newport. If you know both of us, that sentence should make you deeply uncomfortable.

Newport built his career questioning whether we need smartphones, social media, or half the productivity tools the rest of us take for granted. I've spent twenty years implementing the technologies he warns about. We don't agree on much. Which is why you should pay attention when we suddenly agree on AI.

Last week, Newport published a piece using the Jevons Paradox to explain why AI won't eliminate jobs. I'd made essentially the same argument days earlier, from entirely different reasoning. When the optimist and the skeptic arrive at the same destination through different paths, the destination is probably real.

Here's where it gets interesting: we also agree there's a dark side coming. We just disagree on whether you can do anything about it.

The Efficiency Trap Nobody's Discussing

The Jevons Paradox is an 1865 observation about coal. When engineers made steam engines more efficient, coal consumption didn't drop—it exploded. Efficiency made coal so useful that demand outstripped the savings. More efficient engines meant more trains, more factories, more applications nobody had considered before.

Newport applies this to AI: if one programmer can do the work of five, companies won't fire four programmers. They'll build five times more software. The efficiency unlocks demand that was previously constrained. I've watched this exact pattern play out with cloud computing, with automation, with every infrastructure shift that made something cheaper and faster.

When AI makes knowledge work radically more efficient, we won't do less of it—we'll drown in exponentially more of it.

That's the part where Cal and I agree. Your job probably survives. The nature of the work is what's about to get weird.

The Dark Side Is Real (But Not Inevitable)

Newport's concern is what he calls the "dark side" of efficiency gains. Steam engines brought coal pollution that choked cities. Email brought the always-on inbox that colonized evenings and weekends. Every time we make something more efficient, we generate second-order consequences nobody saw coming.

His prediction: AI efficiency will create its own version of constant interruption. If you can generate a marketing campaign in ten minutes instead of ten days, you'll be asked to generate ten campaigns. The tool that was supposed to free up your time will fill it with exponentially more iterations, more requests, more "quick asks" that compound into chaos.

I was on a call last month with a creative director who's already living this. Her team adopted AI tools for layout variations. Three months later, client revision rounds went from two to eleven. Why? Because when revisions are fast and cheap, clients stop thinking carefully about what they want. They iterate their way to clarity on your dime.

This is where Cal and I split.

Newport treats this dark side as almost inevitable—a structural feature of efficiency gains that we'll struggle to contain. I think he's half right. The pressure is real. The outcome isn't predetermined.

The Pattern: Disruption Doesn't Hit Everyone Equally

Here's what I've learned surviving four technology disruption cycles: the same innovation creates completely different outcomes for different people. Not because of luck. Because of skill.

Email didn't ruin every knowledge worker equally. Some people drowned in their inbox, responding to every message within minutes, letting other people's urgency dictate their entire day. Others built filters, set boundaries, learned when to batch responses and when immediacy actually mattered. Same tool, same industry, opposite outcomes—separated by judgment.

I watched the same split with smartphones. Some executives became tethered to Slack notifications at their kid's soccer game. Others turned off badges, set communication hours, and trained their teams that "urgent" had an actual definition. The technology created the pressure. Human judgment determined whether that pressure became productivity or pathology.

The dark side of every efficiency gain isn't a feature of the technology. It's a skill gap. And skill gaps close—for the people who bother closing them.

What the Efficiency Explosion Actually Requires

When AI makes your work five times faster, you're not getting five times more thinking time. You're getting five times more requests. The bottleneck shifts from execution speed to judgment about what's worth executing.

That's the skill almost nobody is building.

I can generate a contract in minutes now. The question isn't "can I do it faster?" It's "should I even be doing this version?" I can produce a financial model in a fraction of the time. The constraint isn't modeling speed—it's knowing which assumptions matter and which are theater.

The professionals who survive aren't the ones who get faster at producing. They're the ones who get better at deciding what's worth producing.

This is uncomfortable because judgment is harder to systematize than execution. You can teach someone to build a financial model in a week. Teaching them when to push back on the assumptions behind it takes years. AI collapses the execution timeline while making the judgment gap more visible and more consequential.

The Question You're Avoiding

Here's the part neither the optimists nor the skeptics want to examine too closely: this isn't a problem we can solve with better tools or clearer processes. You can't automate your way out of needing judgment.

When revision rounds go from two to eleven because clients can iterate cheaply, the answer isn't a better AI tool. It's the skill to say "we're iterating our way to mediocrity—let's define what good looks like before round three." When you can generate five versions of a strategy in an hour, the answer isn't generating a sixth version. It's knowing when more options make the decision harder, not clearer.

That creative director I mentioned? Her solution wasn't technical. She changed the contract. Clients now get three AI-assisted revision rounds, then additional rounds are billed at the old hourly rate. Not because the work got more expensive. Because unlimited iterations were training clients to outsource their own strategic thinking.

The dark side Cal warns about is real. I've seen it eat teams who treated AI as a speed boost without building the judgment to manage what speed unlocks. But I've also seen teams navigate it—not by rejecting the efficiency, but by developing the skill to handle what efficiency makes possible.

What Monday Morning Looks Like

Cal Newport and I agree your job probably survives. We agree there's a dark side to the efficiency explosion. Where we split is whether you're stuck with the consequences or capable of navigating them.

I don't think this is inevitable. I think it's a skill gap. And I think the gap is widening fast between people building judgment and people just building speed.

Here's what to actually do: Pick one place where AI made your work faster in the last month. Now ask—did that speed give you more thinking time, or did it just increase the volume of requests? If it's the latter, the bottleneck isn't your tools. It's your boundaries.

The efficiency explosion is coming whether you're ready or not. The question isn't whether AI eliminates your job. It's whether you're building the judgment to handle five times the volume with the same hours in the day—or whether you're waiting for someone to announce that the chaos has stabilized and it's safe to develop an opinion.

Nobody gets fired the day the AI arrives. They just slowly drown in the exponentially expanding requests that cheap efficiency makes possible.

Are you building filters, or hoping the flood slows down on its own?