The Cognitive Debt of AI: Are Project Managers Getting Worse at Their Own Craft?

The Cognitive Debt of AI: Are Project Managers Getting Worse at Their Own Craft?

By Christopher Scordo, PMP, ITIL · Last updated: August 5, 2026

Quick Answer

Cognitive debt is the slow erosion of critical thinking, memory, and creativity that builds up when we let AI do our thinking for us. Researchers at the MIT Media Lab coined the term after watching people who wrote with an AI assistant show weaker brain activity and struggle to recall what they had just produced. For project managers, whose work runs on judgment, estimation, and reading a room, the risk is real: better deliverables today can quietly cost you sharper instincts tomorrow. The fix is not to abandon AI. It is to keep doing some of the thinking yourself, on purpose.

Introduction

Talk to enough project managers and you hear a worry that few will say out loud in a stand-up: that they are getting a little worse at their own craft the more they lean on AI. The status report writes itself. The risk register fills in seconds. The tricky stakeholder email lands in the right tone on the first try. The work is easier, and the output looks better than ever.

And still, underneath it all, a quiet question. If the tool is doing the thinking, what is happening to mine?

That question is not paranoia. It has a name now, and a growing body of research behind it. After nearly two decades preparing project managers for the toughest moments of their careers, we think it is worth naming plainly, because the professionals who understand the cost are the ones best positioned to avoid it.

What is cognitive debt, and why does AI create it?

Psychologists call the underlying behavior cognitive offloading: handing a mental task to an external tool so you do not have to hold it in your own head. Researchers Evan Risko and Sam Gilbert defined the term in 2016. Writing a phone number down instead of memorizing it is cognitive offloading. So is letting a model draft your project charter.

Offloading is not new, and it is not inherently bad. What is new is how much we can now hand off, and how well the tool hides the cost. In 2025, researchers at the MIT Media Lab gave the pattern a sharper name: cognitive debt. In their study, participants wrote essays in three groups, one using an AI assistant, one using a search engine, and one using only their own brain. The brain-only writers showed the strongest, most widespread neural activity. The AI group showed the weakest. More striking, over 80 percent of the AI users could not quote a single line from the essay they had just finished. They had produced the work without ever really owning it.

A separate 2025 study from the Swiss Business School, published in the journal Societies, put a number on the relationship. Across 666 participants, researcher Michael Gerlich found a strong negative correlation, r = minus 0.68, between frequent AI tool use and critical thinking, driven largely by cognitive offloading. The heaviest users, and the youngest, scored the lowest.

And this is not just a student problem. When Microsoft and Carnegie Mellon surveyed 319 knowledge workers who use AI at work, most said they put less mental effort into the tasks they handed to the tool, and the more they trusted the AI, the less critical thinking they applied. The people who scrutinized the output most closely were the ones with the most confidence in their own expertise.

Three independent studies, three different methods, one direction of travel.

The evidence is converging

Three independent 2025 studies point the same way on AI and thinking

r = −0.68

Strength of the negative link between frequent AI use and critical thinking across 666 people (Gerlich, Societies, 2025).

72%

Of AI-using knowledge workers reported putting less mental effort into analytical tasks they handed to the tool (Microsoft and Carnegie Mellon, 2025).

80%+

Of AI-assisted writers could not quote their own just-finished essay (MIT Media Lab, 2025).

Why are project managers especially exposed?

Most of a project manager's real value is invisible in the deliverable. It lives in judgment. The estimate that feels too optimistic. The dependency nobody flagged. The read on a sponsor who says yes but means not yet. These are pattern-recognition skills, built over years of running projects and being wrong often enough to get calibrated.

Those are exactly the muscles cognitive offloading weakens. You build estimation judgment by estimating, comparing your guess to what actually happened, and adjusting. If the tool produces the estimate and you approve it, you skip the loop that made you good in the first place. You build stakeholder instinct by sitting in the discomfort of a hard conversation, not by having AI soften the email for you. The skills that make a project manager hard to replace are precisely the ones that fade fastest when they are outsourced.

There is an adoption angle too. AI in project work is already mainstream. Roughly one in five project professionals now uses generative AI in more than half of their projects. The tools are not going away, and they should not. But wide adoption means the exposure is wide as well.

Meanwhile the market is moving in the opposite direction from atrophy. The World Economic Forum's Future of Jobs Report 2025 still ranks analytical and creative thinking among the skills employers value most. The capabilities most at risk from over-reliance are the same ones the market will pay the most to keep.

The tradeoff nobody clocks: better output today, weaker instincts tomorrow

Here is the trap, and it is a quiet one. Cognitive debt does not show up on today's project. Today's status report is cleaner. Today's plan looks more thorough. Every individual decision to reach for the tool is rational and produces a better immediate result.

The cost is deferred, which is exactly why it is easy to miss. Debt compounds. Each time you let the model do the thinking on a task you used to do yourself, you get a slightly better artifact and a slightly weaker instinct. Run that for a year and the artifacts are still fine. The instinct is not. Then a project goes sideways in a way the tool did not see coming, a political shift, a novel risk, an estimate that needed a gut check, and the judgment you would have leaned on has quietly thinned.

The MIT researchers found something telling on this point. When people who had been using AI switched to working unaided, their brain activity did not bounce back to the level of those who had been thinking for themselves all along. The muscle does not simply re-engage because you set the tool down.

The project manager's judgment muscles: what weakens them, how to keep them alive

Judgment muscle	What weakens it	How to keep it alive
Estimation	Approving AI estimates without forming your own.	Guess first, then compare to the tool and to actuals.
Risk identification	Letting the model populate the risk register for you.	Draft your own risks, then ask AI what you missed.
Reading stakeholders	Outsourcing hard conversations to polished AI drafts.	Own the framing yourself; use AI only to refine tone.
Creative problem-solving	Accepting the first answer the tool produces.	Generate your own options before you prompt, then compare.

How to keep the muscle alive

The answer is not to swear off AI. Used well, it is one of the best productivity gains the profession has seen in years. The answer is to be deliberate about keeping some thinking manual, on purpose, so the tool amplifies your judgment instead of replacing it.

A few practices that work:

• Think first, then prompt. Draft your own estimate, risk list, or stakeholder approach before you open the tool. Then use AI to pressure-test it, not to produce it. You keep the reps, and the tool sharpens them.
• Make the AI show its work, and check it. Ask for the reasoning, not just the answer. The knowledge workers who stayed sharp in the research were the ones who challenged the output rather than accepting it.
• Protect a few high-judgment tasks as manual by default. Estimation, risk identification, and hard stakeholder conversations are where your instincts are built. Keep doing enough of them yourself to stay calibrated.
• Treat AI output as a first draft, never a final answer. In the Microsoft research, the critical thinking that held up was the kind aimed at checking and correcting the tool's work. Make that review real, not a rubber stamp.

None of this is anti-technology. It is the same discipline good project managers already apply to any powerful tool: know what it is for, know what it costs, and stay in control of the decision. If you want a broader view of which capabilities are rising and which endure, our look at the top project manager skills for 2026 is a useful companion to this piece.

The project managers who come out ahead will not be the ones who use AI the most, or the least. They will be the ones who use it without going into cognitive debt, who let it carry the load while keeping their own judgment sharp enough to know when it is wrong.

Frequently Asked Questions

What is cognitive debt?

Cognitive debt is the gradual erosion of critical thinking, memory, and creativity that builds up when you let AI do your thinking for you. Researchers at the MIT Media Lab coined the term in 2025 after finding that people who wrote with an AI assistant showed weaker brain activity and often could not recall the work they had just produced.

Does using AI actually make you worse at thinking?

Heavy, unexamined reliance can. A 2025 study of 666 people found a strong negative correlation between frequent AI use and critical thinking, driven by cognitive offloading. A separate Microsoft and Carnegie Mellon survey of 319 knowledge workers found most reported putting less mental effort into the tasks they handed to AI. The effect is tied to over-reliance, not to using the tools at all.

Why are project managers especially at risk?

Most of a project manager's value lives in judgment: estimation, risk sense, and reading stakeholders. Those are pattern-recognition skills built through repetition, and they are exactly the muscles that weaken when the thinking is handed to a tool. That makes project managers unusually exposed to cognitive debt.

How can I use AI without losing my critical thinking?

Do some of the thinking first, on purpose. Draft your own estimate or risk list before you open the tool, then use AI to pressure-test it. Ask the model to show its reasoning and check it. Keep a few high-judgment tasks manual by default, and treat AI output as a first draft rather than a final answer.

The Bottom Line

AI is not making project managers worse at their craft. Using it without thinking is. Cognitive debt is real, it is measurable, and it compounds quietly while the deliverables still look good. The professionals who stay sharp will be the ones who treat their own judgment as an asset worth maintaining, and who keep exercising it even when the tool could do the work for them.

About the Author

Christopher Scordo, PMP, ITIL is Founder and Managing Director of PMTraining, a PMI Premier Authorized Training Partner that has trained more than 150,000 professionals over 19+ years. He is the author of multiple best-selling PMP exam prep books and a long-standing member of PMI. Read his full bio on PMTraining.com.

Sources:

Kosmyna et al., Your Brain on ChatGPT: Accumulation of Cognitive Debt When Using an AI Assistant for Essay Writing, MIT Media Lab (2025)

Gerlich, M. AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking, Societies (2025)

Lee et al., The Impact of Generative AI on Critical Thinking, Microsoft Research and Carnegie Mellon University, CHI 2025

Risko, E. F. and Gilbert, S. J. Cognitive Offloading, Trends in Cognitive Sciences (2016)

World Economic Forum: The Future of Jobs Report 2025

PMI: Shaping the Future of Project Management With AI