Education Can't Keep Up

About ten years ago, I interviewed one of the brightest juniors I'd ever met.

Sharp. Eager. The kind of person who asked good questions and listened to the answers. We hired him straight out of university, knowing he'd need months before he could contribute to real projects. That was the deal back then. Every software company in Serbia understood it—graduates came with theory and left the practical skills for us to teach.

So I taught him. Not just me—my senior developers too. We structured a learning path for him. Gave him real projects with training wheels on. Sat with him when he got stuck. Explained not just the code, but the people skills—how to talk to clients, how to collaborate with a team, how to handle the moment when everything breaks at 5 PM on a Friday.

For six months, he cost us more than he produced. We knew that going in. We swallowed it. By month nine, he was starting to carry his own weight. By month twelve, he was genuinely good.

Then he left.

Found a job paying three to four times what we offered. I couldn't blame him. I would have done the same thing at his age.

Then it happened again. And again. Over the next two years, the pattern repeated enough times that I stopped counting and started restructuring our contracts.

But here's what took me years to articulate: The problem was never the juniors. They did exactly what rational people do. The problem was that the education system produced people who needed another six to twelve months of real-world training before they could do real work.

I was subsidizing a broken system with my own time, my team's energy, and my company's money.

Scaling What I Learned the Hard Way

That experience didn't make me bitter. It made me restless.

If I could only train a handful of juniors inside Orange Hill—and lose half of them before seeing any return—then maybe the answer was to scale the knowledge differently. Reach more people. Build something bigger than one company's onboarding program.

That's one of the reasons my colleague Milos and I started the IT Serbia Podcast—the first IT podcast in Serbia. We covered software craftsmanship, career development, the gap between what universities taught and what companies needed.

Separately, I threw myself into the software development community. Orange Hill and I organized conferences, spoke at meetups, recorded videos. Anything to push the conversation forward.

Soon, private coding schools started appearing across Serbia. Five to ten years ago, they were everywhere—bridging the gap that universities couldn't close fast enough. It worked. The candidates we saw in 2018 were better prepared than the ones in 2012. The ecosystem adapted.

That was the old gap. Measured in years. Fixable with effort.

The new gap is something else entirely.

I'm Watching the Same Movie

Fifteen years later, the pattern is identical. Except the people who are stuck aren't 22-year-old graduates.

They're CEOs.

I've had conversations with C-level executives over the past year that felt like time travel. They remind me of those juniors—smart, motivated, aware that something fundamental has shifted, but unable to close the gap between knowing and doing.

They see the potential. They've read the headlines. Some have experimented personally—played with ChatGPT, maybe tried Claude. A few have even built personal workflows that save them time.

But when it comes to rolling AI across their organizations? Stuck.

The same way my juniors were stuck. Not because they lacked intelligence. Because the education available to them didn't match the reality they needed to operate in.

Why the Gap Is Exponentially Worse This Time

When I was hiring juniors in 2012, the education lag was maybe two to three years. Universities covered the fundamentals but skipped the patterns and frameworks that mattered in production. Fixable. Tedious, but fixable.

AI doesn't move like that.

AI capabilities change monthly. Sometimes weekly. The gap between what a course can teach and what the technology can do isn't measured in semesters anymore—it's measured in model releases. By the time a curriculum committee approves a new AI module, the tool it's based on has been superseded twice.

PwC surveyed 4,454 CEOs across 95 countries in January 2026. Fifty-six percent reported no financial benefit from their AI investments—neither revenue gains nor cost savings. A separate NBER study of 6,000 executives found that nearly 90 percent of firms report AI has had no impact on productivity over the past three years.

Billions invested. Almost nothing to show for it.

An economist at Apollo looked at the data and said what everyone was thinking: "AI is everywhere except in the incoming macroeconomic data." It's Robert Solow's 1987 paradox all over again—"You can see the computer age everywhere but in the productivity statistics." History doesn't repeat, but it rhymes.

Wikipedia

“"You can see the computer age everywhere but in the productivity statistics."”

— Robert Solow, Nobel Laureate, 1987

The difference is speed. The computer age took a decade to resolve its productivity paradox. AI won't wait that long. The companies that figure out the education problem first will compound their advantage exponentially. The ones that don't will fall behind at the same rate.

Meanwhile, at the Bottom of the Ladder

While C-levels struggle to deploy AI across their organizations, the people who used to enter this industry are being shut out.

I wrote about this in A Letter to Juniors—employment for software developers aged 22 to 25 has declined nearly 20 percent since late 2022. A Harvard study covering 285,000 firms found that when companies adopt generative AI, junior employment drops nine to 10 percent within six quarters. Senior employment barely changes.

Computer science graduates now face 6.1 percent unemployment—higher than liberal arts graduates. Seventy percent of hiring managers say AI can perform intern-level work.

The juniors can't get hired because AI made seniors more productive. The seniors can't deploy AI effectively because nobody taught them how. And the education system—the thing that was supposed to prepare both groups—is watching from the sidelines, updating curricula that were outdated before the ink dried.

It's the same structural failure I saw 15 years ago. But this time, it's hitting both ends of the org chart simultaneously.

Why Generic Courses Are a Dead End

Here's what I keep hearing from the C-levels I talk to: "We've sent people to AI courses. It didn't stick."

Of course it didn't.

It's the same reason university didn't prepare my juniors. Generic courses teach you how a tool works. They don't teach you how to use it in your specific context, with your specific data, your specific workflows, your specific team dynamics.

A CEO doesn't need to understand how Claude Code works in the abstract. She needs to understand how it fits into her company's sales process. How it connects to their CRM. How it changes the way her marketing team produces campaigns. How to set up boundaries so her developers' Git repositories don't get corrupted by well-meaning executives experimenting with code generation.

That's not a course. That's a consulting engagement.

McKinsey found that C-suite leaders are more than twice as likely to blame employee readiness as a barrier to AI adoption than to blame their own role. But Deloitte's data tells a different story—only 20 percent of companies rate their talent as "highly prepared" for AI, while education was the number one way companies adjusted their talent strategies. They know the problem. They just don't have the right solution.

And then there's the phenomenon that nobody wants to talk about.

Fifty-three percent of C-suite leaders hide their AI use at work to avoid judgment. The highest concealment rate of any group. Meanwhile, only 7.5 percent of employees receive extensive AI training.

The leaders who should be championing AI adoption are secretly using it at their desks while providing zero guidance to their teams. They're ashamed of needing AI—and that shame is paralyzing their organizations.

I wrote in Personally Productive, Organizationally Stuck about the three levels of AI maturity—personal mastery, deliberate self-education, and organizational rollout. Most leaders stall between levels one and two. They've experienced the transformation personally, but they can't replicate it for their teams because the gap between personal use and organizational deployment requires a fundamentally different kind of knowledge.

Generic courses won't give them that knowledge. The same way university never gave my juniors the practical skills they needed.

The Bridge That Works

So what worked fifteen years ago?

Not the universities—they were too slow. Not the generic courses—they were too broad. What worked was practitioners teaching practitioners. People who had solved the problems themselves, sharing that knowledge directly, hands-on, customized to the person sitting across from them.

That's what the coding schools did when they were good. That's what I did on the podcast. That's what happened at every meetup where a senior developer sat with a junior and said, "Here's how this works in the real world."

The same principle applies now—but at a higher level, with higher stakes, and for a different audience.

I wrote in Stop Presenting. Start Building. about the shift from slide-deck consulting to prototype-first consulting. The same logic applies to AI education. You can't teach an executive to use AI by lecturing them. You teach them by sitting next to them, understanding their specific business, and building something real—together.

A consultancy can do what no course can:

Understand the complex business needs of each company—not just "how to use AI" but where AI creates the most value in this specific organization.

Map AI opportunities across teams, spotting inefficiencies that span departments—things a company might not even notice because they've been doing it that way for ten years.

Create a prioritized list of interventions—not everything at once, but the changes with the highest impact first.

Build the solutions and teach the employees to maintain and improve them. That last part is the real value. Not dependency on a consultant. Capability transfer that compounds over time.

“"Successful AI implementations pick one pain point, execute well, and partner smartly."”

— MIT NANDA Initiative, 2025

The MIT researchers who studied why 95 percent of AI pilots fail found that the five percent that succeed share three traits: narrow focus on a specific pain point, deep collaboration between AI teams and end users, and measurable outcomes tied to business results. Not broad ambition. Not generic training. Precision.

That's what I learned the hard way with my juniors, too. The ones who succeeded weren't the ones who attended the most courses. They were the ones who got hands-on mentorship, tailored to their specific gaps, from someone who had already walked the path.

The Same Optimism, Fifteen Years Later

I started this piece with a story about losing juniors I'd invested in. About contracts I had to restructure. About a system that took my time and gave me nothing back.

But that's not the whole story.

That frustration pushed me to start a podcast. To join a community. To speak at events, record videos, help build an ecosystem that—imperfectly, slowly, but genuinely—closed the gap for thousands of developers across Serbia.

The heartbreak didn't make me cynical. It made me find a better way.

I'm in the same place now. The gap is bigger. The stakes are higher. The people who are stuck aren't 22-year-olds wondering how to land their first job—they're leaders wondering how to transform their organizations before the market leaves them behind.

But the pattern is the same. Education can't keep up. It never could. The bridge has always been practitioners who care enough to teach what they know—not in a lecture hall, not in a generic webinar, but sitting next to you, understanding your specific reality, and helping you build something that works.

That's what we do at Orange Hill. Not because I read it in a textbook. Because I've been building that bridge for fifteen years—and the other side is worth getting to.

Tihomir Opačić is the founder of Orange Hill, where he helps organizations bridge the gap between AI potential and AI reality. If your team is stuck between knowing AI matters and making it work, reach out.