The Great Leveling

For two years, the dominant story about AI has been about who it will replace. There’s a quieter, more interesting story buried in the research — one about who it will elevate. When Boston Consulting Group ran 758 of its consultants through a controlled study with GPT-4, the below-average performers improved by 43 percent. The above-average performers improved by 17 percent. AI didn’t widen the gap between strong and weak. It compressed it. And if that finding holds at scale, the implications for global talent are far more profound than any layoff headline.

The 43 percent story

The BCG result isn’t a one-off. Erik Brynjolfsson and his colleagues at Stanford ran a similar experiment on 5,179 customer service agents at a Fortune 500 company. The result was almost identical in shape: novices saw a 34 percent productivity gain, while the most experienced agents barely moved the needle. As Brynjolfsson put it, AI “disseminates the best practices of more able workers and helps newer workers move down the experience curve.”

Translate that into plain English: AI is acting as a portable mentor. The thing that used to take a decade of on-the-job apprenticeship — the tacit knowledge, the pattern recognition, the instinct for what “good” looks like — can now be partially loaded into a junior employee’s workflow on day one.

This is a genuine inversion. MIT Sloan researchers note that for the past 30 years, every wave of digital technology widened the gap between high-skilled and low-skilled workers. AI is the first technology in a generation that appears to do the opposite.

Work is moving from execution to judgment

The T2 Conline piece that prompted this column puts the shift cleanly: competition is moving from execution to judgement. From who can do the work to who can direct it. That framing matters because it reframes the entire “AI will steal your job” panic. If skill compression is real, then the bottleneck for most knowledge work is no longer talent. It’s taste, judgment, and the ability to ask the right question.

And here’s the part the doomsayers keep missing: judgment is teachable, learnable, and — crucially — not geographically constrained. A 24-year-old in Manila with strong critical thinking and a good prompt can now produce work that rivals what a mid-level professional in London produced five years ago. That isn’t a hypothetical. That’s what 43 percent improvement looks like in practice.

Of course, there’s a catch. The same BCG study identified what the researchers called a “jagged technological frontier.” Inside the frontier, AI helps. Outside it, AI produces confident nonsense and can actually make workers worse. The expertise that survives is the expertise required to know which side of the line you’re on. That’s a skill, but it’s a different skill from the one most professionals have spent their careers building.

The borderless bonus

This is where the future-of-work story gets genuinely exciting. For the better part of three decades, the global outsourcing industry has been built on a simple proposition: capable people in lower-cost geographies can do work that capable people in higher-cost geographies used to monopolize. AI doesn’t threaten that model. It supercharges it.

Consider the implications. AI-related jobs are growing faster in middle-income countries than in high-income ones. The Philippines, India, and Vietnam — long the engine rooms of global services delivery — are now home to a workforce that can punch dramatically above its experience level when paired with the right tools. The premium that wealthy economies once charged for “having access to senior talent” is eroding. Not because senior talent is being replaced, but because the skill ceiling for everyone else is being raised.

For the first time, the question business leaders should be asking isn’t “how do we automate this role?” It’s “where in the world can we now find the person who will do this role brilliantly with AI in hand?” Geography becomes a sourcing decision, not a quality constraint. The shift from where you sit to what you deliver — already a decade in motion thanks to remote work — just acquired a turbocharger.

The honest caveat

None of this is frictionless. Entry-level hiring has genuinely cratered in some sectors — software job postings for graduates have fallen sharply, and the bottom rung of the corporate ladder is getting structurally rebuilt. The risk is real: if companies use AI to thin out their junior ranks rather than expand them, the long-term pipeline of judgment-capable mid-careers dries up. That’s a problem AI cannot solve, because judgment, in the end, still has to come from somewhere.

But the larger point stands. The default assumption that AI is a vacuum cleaner pulling jobs out of the economy is being quietly contradicted by the data. What it actually appears to be is a rising tide that lifts the less skilled faster than the highly skilled — the first time in a generation a technology has done that. If you believe in opportunity, that’s not a threat. It’s the most democratizing shift in knowledge work since the laptop.

The companies that win this decade won’t be the ones that fired the most people. They’ll be the ones that figured out how to find capable, judgment-driven talent anywhere on Earth and hand them the tools to perform like specialists. The flattening is real. The question is who’s standing on the high ground when it finishes.

The question for your business

Are you using AI to shrink your team — or to widen the talent pool you can draw from?

Read more thought leadership articles here:

• • Robots Need Colleagues Too
  • The Philippines Trained AI. Now AI Is Coming For It.