A 2025 MIT study reveals the cognitive risks of outsourcing critical thinking to AI. Learn why executives must protect analytical reasoning to maintain leadership.
A chief operating officer sat in silence as her team presented a flawless, machine-generated strategic plan. The formatting was impeccable, but the foundational assumptions were critically flawed. Not a single person in the room could explain how the underlying conclusions were reached. This scenario reflects a growing statistical reality across global business, where half of surveyed leaders report that their organizations are already experiencing AI-related de-skilling.
This fundamental shift in how professionals process information requires urgent attention. Companies are rapidly adopting powerful new software to drive efficiency, but they are inadvertently removing the necessary friction that builds leadership capacity. When complex analysis is entirely outsourced, the individuals responsible for final oversight slowly lose their ability to evaluate the work. We must understand exactly what happens to human capability when algorithms perform the heavy lifting.
According to an article by Daniel Goleman and Elizabeth Solomon for Korn Ferry, a 2025 MIT study examined how habitual digital assistance impacts human cognition. The full experiment involved 54 participants in the first three sessions, with 18 people assigned to each of the three groups. These participants wrote essays under specific conditions using either ChatGPT, Google Search, or no digital tools. Researchers then measured brain activity, memory recall, essay quality, and the participants' sense of ownership over the final product.
The findings reveal a distinct separation in neural behavior during the task. In the comparison described by Korn Ferry, participants writing without digital tools showed the strongest and most widespread brain connectivity. Conversely, the ChatGPT group showed the weakest connectivity and had greater difficulty remembering what they had written. Korn Ferry cautions that this preliminary study is not robust enough for broad conclusions, yet it raises critical questions about human capability.
This research does not establish permanent brain damage or irreversible decline. Instead, it measures differences in neural connectivity and memory performance during a specific writing task. The term cognitive atrophy serves as a powerful leadership metaphor rather than a clinical diagnosis established by the experiment. It describes the potential loss of mental sharpness that occurs when people stop practicing the difficult forms of thinking that algorithms can now simulate.
For executives, these findings translate into severe boardroom liabilities. BCG-related reporting states that half of surveyed leaders observe de-skilling, while more than 60 percent expect it to become a material threat within three to five years. The capabilities identified as most vulnerable are the exact skills required for leadership. These include problem framing, judgment, creative thinking, analysis, and solution generation.
The risk becomes even more apparent when we consider the demands of the modern labor market. The World Economic Forum reported that seven in ten companies considered analytical thinking essential in 2025. The same report identifies analytical thinking, resilience, and leadership as critical skills in the changing economy. Organizations face a stark tension between deploying software to increase speed and preserving the human judgment required to supervise highly consequential systems.
If a team cannot critically evaluate a machine-generated strategy, the company absorbs enormous operational risk. A poorly designed workflow may make an individual faster today while weakening the person's ability to recognize a bad premise tomorrow. This structural vulnerability forces business leaders to rethink how they evaluate digital initiatives. Speed and throughput are no longer the only metrics that matter when long-term organizational competence is at stake.
A 2025 study by Microsoft Research and Carnegie Mellon University adds further clarity to this dynamic. Researchers surveyed 319 knowledge workers across 936 reported examples of generative-AI use at work. They found an association between greater confidence in the system and less reported critical-thinking effort. However, greater confidence in a person's own task-specific ability was associated with more critical-thinking effort.
This suggests that artificial intelligence does not necessarily eliminate human thought, but it relocates it. The Microsoft and Carnegie Mellon researchers reported that assisted work shifted critical thinking toward verification, response integration, and task stewardship. The immediate cost to an organization occurs when employees skip this critical verification step. A workflow that relies entirely on automation may produce faster results today while destroying the team's ability to challenge an attractive answer tomorrow.
The research accounts for specific behavioral variables that dictate how individuals interact with digital tools. In the Microsoft and Carnegie Mellon study, the primary variable affecting cognitive effort was the user's level of confidence. Greater confidence in the artificial intelligence system correlated directly with less reported critical-thinking effort from the knowledge worker. When employees trusted the software implicitly, they were far more likely to accept outputs without verifying the underlying logic.
Conversely, the data revealed a protective variable in human behavior. Greater confidence in a person's own task-specific ability was associated with more critical-thinking effort. Experienced professionals who deeply understood their domain used the technology as a challenger rather than a replacement for their own reasoning. This variable highlights why experienced executives interact with these tools differently than junior employees who lack foundational industry knowledge.
I remember landing at Heathrow after a brutal overnight flight from New York. I had a board meeting in three hours. The standard advice of getting eight hours of sleep felt like a cruel joke. That was the exact moment I realized our readers do not need perfect scenarios.
They need triage protocols. They need to know what the science says about recovering cognitive function when you only managed three hours of terrible sleep at high altitude. When you step into a high-stakes room under that level of fatigue, your brain relies entirely on deep analytical habits built over years of unaided struggle. This is where cognitive performance and mental clarity become non-negotiable.
If you have outsourced your toughest problem-solving to an algorithm, you lack the mental reserve required to navigate ambiguity under pressure. Korn Ferry compares this cognitive effort to resistance training. People lift weights because the resistance itself develops strength, and the same principle applies to leadership wisdom. Leaders accumulate judgment through lived experience, writing difficult proposals, and experiencing the consequences of decisions.
Leaders must evaluate digital initiatives by asking what capabilities their future organization will require. Korn Ferry explicitly distinguishes between using tools to avoid thinking and using them to spark curiosity or locate evidence. Routine administrative tasks offer little developmental value and are perfect candidates for automation. However, executives must actively protect the friction that develops judgment and originality.
The first operational shift is to reserve problem framing entirely for humans. A useful operating rule is that humans frame the question while the algorithm helps test the answer. Before consulting an external tool, the decision owner should write down the problem, the core assumptions, and the leading alternative explanations. This practice forces the brain to engage in the necessary struggle that builds analytical strength.
Organizations should also separate generation from evaluation during consequential projects. A human should create the initial thesis before a machine generates counterarguments or missing considerations. The human then verifies the evidence and revises the thesis accordingly. This specific sequence preserves the essential verification and oversight functions highlighted in the Microsoft and Carnegie Mellon research.
Companies must avoid heavily automated onboarding for junior talent. If early-career employees use software to write every first draft, they lose the precise experiences through which business judgment develops. Young professionals need to periodically complete unaided exercises and verbally defend their reasoning. This ensures they build the internal competence required for long-term executive performance.
Because higher confidence in these systems associates with less critical thinking, leaders must treat blind trust as a governance failure. Teams should be trained to calibrate their confidence against domain knowledge, evidence quality, and the consequences of being wrong. Final accountability must remain entirely human for decisions involving capital allocation, safety, reputation, and long-term strategy. The human decision-maker must always be able to explain the recommendation without relying on the software's authority.
Not every task should be automated from start to finish. Leaders can preserve useful resistance by asking teams to draft a recommendation before opening any software. They should also require employees to periodically solve selected problems without digital assistance, debate competing explanations, and defend decisions verbally. These deliberate practices follow the argument from Korn Ferry that some struggle is developmental rather than wasteful.
The strongest version of this argument is not a fear-driven narrative about technology making people less intelligent. It is a precise warning about the compounding value of deliberate practice. Small daily choices about when to struggle with a problem and when to automate it accumulate over decades. Leaders who consistently engage in rigorous, unaided analysis build a compounding reservoir of strategic intuition.
Those who default to algorithmic answers risk plateauing their own intellectual development. The goal is to build workflows that remove low-value friction while preserving the productive resistance that forces human adaptation. This balanced approach to focus and cognition ensures that your most valuable asset remains sharp. Over time, the leaders who maintain their capacity for original synthesis will vastly outperform those who only know how to prompt a machine.
Start protecting your mental capacity by changing how you begin your most difficult tasks. Write out your core assumptions and initial thesis on a blank sheet of paper before you consult a single digital tool today.
Stay connected for research and practical guidance on executive performance, energy, focus, sleep, recovery and longevity. Ideas built for people who want to stay sharp, capable and effective for the long run.
Build habits and systems that support clear thinking, steady energy and long term capacity throughout a demanding career.
explore the Blog
