Michael Forrester

Every conversation that I am having with C-levels, Engineering Managers, and L&D about AI right now seems to start with the same premise. Because of AI, we’ll need fewer people. And every conversation that follows is some variation on how to manage the consequences of that premise. Slow hiring. Cut juniors. Consolidate roles. Maybe sprinkle some training on top so it feels less brutal. Then frame the whole thing as investing in AI and bringing people along.

I think we’ve got this backwards, and I want to explain why.

Start With the Obvious Question

Who is going to implement all this AI?

The infrastructure has to be stood up by people. The third-party SaaS platforms have to be integrated by people. The abstraction layers have to be selected, configured, and stitched into existing systems by people. Even if you go custom and build your own frameworks from scratch, that’s people. The data pipelines, the security reviews, the model evaluation, the prompt engineering, the agent orchestration, the governance, the change management, the user enablement — every single one of those is a human capability. Yes I know.. it can be assisted and faciliated by AI, but it won’t be done by AI. Not yet.. not in perhaps even a few years if at all.

So the idea in 2026 and 2027 that you invest in AI and bring people along has the relationship backwards. People aren’t cargo on the AI vehicle. People are the engine. At least for now, and possibly.. for a while.

The Substitution Assumption Is Doing All the Work

The “fewer people” thesis rests on an assumption that almost never gets stated out loud. It assumes AI will eventually match human cognition and judgment well enough to fully substitute for it. That is a bet on a future state, being placed with today’s workforce decisions. But we are remarkably bad at predicting the future completely…. at least in specifics.

And it cuts against how innovation actually happens.

Technology solves human problems. Humans frame those problems. Humans recognize when an answer doesn’t fit reality. Humans carry the institutional context that makes any of the work matter in the first place. Take humans out of the orchestration loop and you don’t just lose your junior pipeline. You lose the source of the questions.

The Data Is Already In, and It Is Brutal

Gartner surveyed 350 executives at billion-dollar enterprises in the third quarter of 2025. About 80 percent reported AI-related workforce reductions, but those cuts showed no correlation with ROI. Companies that cut hardest got the same returns as companies that didn’t.¹

Helen Poitevin, Distinguished VP Analyst at Gartner, put it bluntly: “Many CEOs turn to layoffs to demonstrate quick AI returns; however, this disposition is misplaced. Workforce reductions may create budget room, but they do not create return."¹

Forrester’s “Predictions 2026: The Future of Work” report goes further. The firm predicts that more than half of AI-attributed layoffs will be quietly reversed, and labels much of the current wave AI washing — financially motivated cuts wearing an AI badge.² Their analyst Betsy Summers calls out the same dynamic: companies announcing AI-related layoffs often don’t have mature, vetted AI applications ready to fill those roles.²

Meanwhile, more Forrester-tracked AI investment leaders expect AI to increase headcount over the next year (57 percent) than to decrease it (15 percent).³

The cut isn’t paying off. The reframe is.

Since When Did Businesses Respond to Capability Expansion by Getting Smaller?

When has a business ever responded to a capability expansion by deciding to shrink?

Not with steam. Not with electricity. Not with spreadsheets. Not with the internet.

Every prior efficiency leap expanded what organizations attempted, and demand for human work grew into the space the technology opened up. Economists call this Jevons Paradox — the observation that improvements in efficiency tend to increase total demand for the resource, not reduce it. It’s been the most well-documented pattern in industrial history for 160 years.

I have yet to meet a business that felt it had pursued every opportunity, every market, every product idea it had capability for. Most organizations I work with carry backlogs in the hundreds, sometimes thousands of items. Things they could never staff for. Experiments they could never run. Customer segments they could never serve. Quality work they could never get to.

Now AI lands as a force multiplier capable of two-to-ten times productivity gains, and the response is to shrink? That isn’t strategy. That’s a reflex in a strategy costume.

The Actual Reframe

We are not investing in AI. We are investing in a business whose core capabilities are AI and people working together, and you genuinely cannot have one without the other. That changes everything about what the investment actually looks like.

If your workforce is the core enabler of AI in your business, then upskilling stops being a kindness or a hedge and becomes the actual investment. Re-contextualizing people as the landscape shifts. Familiarizing them with new tools as they emerge. Teaching evolving best practices continuously, not as a one-time event. Re-exposing the workforce to a quickly changing field on a regular cadence.

That is the investment. Everything else is the wrapper.

And None of This Works Without Slack

Here is the part that gets left out of every AI strategy deck I’ve seen. None of this works without slack or some kind of space.

Tom DeMarco wrote a book by that name in 2001, and the argument has aged into prophecy.⁴ His core thesis was that hyper-efficient organizations destroy their own capacity to change because they leave no room for adaptation, learning, or innovation. His exact words:

“What got cut out of the most aggressively purged organizations is the capacity to change. The so-called restructurings have, in most cases, optimized the present steady state, only at the expense of the future."⁴

Read that sentence with AI layoffs in mind. He was describing this moment two decades early because we are still dealing with the business and human psychology around productivity and technology adoption that we were seeing almost three decades ago.

DeMarco’s argument is operational, not sentimental. If we want our people to upskill, adapt, and stay ahead of a technology that evolves on a monthly cycle, we have to give them the breathing room to do it. You cannot run a workforce at 100 percent utilization and also expect it to absorb new tools, develop new judgment, and innovate against new problems. The math does not work.

“Change and reinvention require slack," DeMarco wrote.⁴ “Money is fungible, but people are not."⁴

Slack isn’t a luxury. It is the operating condition under which adaptation becomes possible. Take it away in the name of efficiency, and you have optimized the present steady state at the expense of the future. Exactly as DeMarco warned.

So Why Is “Cut” the Default?

The honest answer, I think, is that cutting is the move executives know how to make quickly, and that the market rewards in the short term. Investing in people is harder, slower, and pays off over a horizon longer than most quarterly earnings calls.

But the data is already telling us which approach is actually working and which is generating regret. The companies treating AI as a capacity expander are pulling ahead. The companies treating it as a substitute are queueing up for the reversal Forrester is forecasting.

We are not investing in AI and bringing people along. We are investing in a business that runs on people who run on AI. That is not a slogan. That is the strategy.

Anything less is a story we are telling ourselves on the way to learning an expensive lesson. One that I professionally am working to unravel both strategically and tactically one conversation at a time because the message is still the same… people are the customer, the problem, the solution, and the answer. As much as we don’t want that to be the cause, AI is going to augment and assist even if it replaces everything we do know, it cannot and will not replace the core of our society, which is for better or for worse… decidely human.

References