The Data Arrived. Most Companies Missed It.

Everyone’s been waiting for the data on AI and jobs.

It arrived. In the last two weeks of March, Goldman Sachs, the Federal Reserve Bank of Atlanta and Harvard Business School all published major findings on what AI is actually doing to workers.

The answer isn’t what the headlines prepared you for.

What the data actually says

Goldman Sachs released its AI Adoption Tracker on April 1. Employees at companies using AI save 40 to 60 minutes a day. 75% say they can now complete tasks they couldn’t do at all before. Academic studies in the report show a 23% average productivity uplift.

The Federal Reserve published a working paper on March 25 surveying nearly 750 corporate executives. The finding that matters: the average impact on employment headcount was close to zero. Companies aren’t shrinking workforces because of AI. They’re reinvesting the productivity gains into expanding capabilities, R&D and upskilling.

Harvard analyzed nearly all U.S. job postings from 2019 through early 2025. Automation-prone roles dropped 13 to 17% after ChatGPT launched. But augmentation-friendly roles grew 20 to 22%. The jobs where AI makes people better at their work are growing faster than the jobs AI threatens.

The Dallas Fed added another layer. Wages are rising in AI-exposed occupations. Not falling. Up 7.5% nationally since fall 2022.

This isn’t one study making a hopeful claim. It’s four independent research institutions converging on the same conclusion in the same window.

AI is making workers more productive. Not replacing them.

Subscribe now

Why this isn’t landing

You’d think this would be front-page news. It mostly isn’t.

The layoff headlines are louder. Oracle. Block. The Challenger data. Those stories have names and numbers and human cost. They travel faster.

But here’s what the layoff coverage misses. The same Goldman Sachs report found that 80% of companies haven’t adopted AI in any meaningful way yet. The organizations cutting headcount and citing AI are a small minority. And as we’ve covered before, a significant number of them admit they’re using AI as framing for decisions that have nothing to do with capability.

The real story isn’t that AI is replacing people at the companies using it. It’s that most companies aren’t using it at all. And the ones that are? They’re pulling away.

What pulling away looks like

It doesn’t look like massive deployments or executive mandates.

It looks like a company called Omnisend giving employees a 2 to 4% raise for demonstrating AI proficiency. Their sales team’s lead follow-up success rate went from 20% to nearly 100% after implementing AI tools. Not because they hired more people. Because the people they had got better.

It looks like FedEx building an AI literacy program across 400,000 employees. Not to cut jobs. To make drivers and customs teams and operations staff more knowledgeable and promotion-ready.

It looks like Citigroup building a network of 4,000 peer AI champions across 84 countries. Not a top-down training mandate. A peer influence model that reached 70% adoption.

It looks like LinkedIn opening its Hack Week to every employee, not just engineers. 3,500 participants. Over half were first-time hackers building things they’d never built before.

None of these stories involve removing people. All of them involve people becoming more capable than they were before.

The adoption gap is the competitive gap

Here’s the structural point most organizations are missing.

The question isn’t whether AI helps workers. That question has been answered. Repeatedly. By credible institutions with large datasets and rigorous methodology.

The question is why 80% of companies are still sitting on the sideline while the evidence says move.

Some of it is fear. The headlines create a climate where AI feels dangerous. Leaders hear “AI layoffs” and assume adoption means headcount reduction. So they freeze. They wait for more clarity. They study the problem instead of building the system.

Some of it is sequence. Organizations try to pick the right tool before they’ve redesigned the workflow. They buy licenses before they’ve built fluency. They announce AI initiatives before they’ve created a culture where experimentation is safe.

And some of it is incentive. In most companies, the people affected by AI decisions have no stake in the outcome. They’re told to adopt. They comply or they resist. Neither produces the kind of honest feedback that makes adoption actually work.

The ownership angle

I work inside an employee-owned company. That changes the math.

When the people using AI are also the people who own the company, the incentive to experiment is real. Not performative. Not compliance-driven. Genuine.

An employee-owner who finds a way to save an hour a day isn’t just more productive. They’re contributing to something they have a stake in. That hour compounds. Not just operationally. Financially.

That’s not a culture initiative. It’s a structural advantage.

The Goldman Sachs data says AI saves 40 to 60 minutes a day for the people who use it. The question is what happens with that time. In most organizations, it disappears into the system. In an ownership structure, it gets reinvested by the same people who created it.

That’s a different kind of adoption. And the data suggests it’s the kind that works.

The window

The evidence is no longer ambiguous. AI makes workers better at their jobs. The organizations acting on that are pulling ahead. The ones waiting are falling behind. And the gap is compounding.

This isn’t a technology problem anymore. It’s a decision quality problem.

The data arrived. The question is whether your organization noticed.