By 2026 most executive workflows run through AI from end to end. Strategy decks are drafted in a window and edited in three. Board memos are summarised before they arrive. Customer research, competitor analysis, and even the first-pass read of a hiring file are AI-mediated. The productivity gains are real and the tools are not going away.
The question for executives is narrower than the one in the headlines. What cognitive acts do you keep doing yourself, every working day, so that the AI integration doesn't hollow out the judgment that defines the role — and so that the way you work models the right cognitive culture for the team under you?
Why this matters — the published evidence
The 2026 reporting on workplace AI fatigue is the freshest reference. Fortune's March 2026 piece on "AI brain fry" summarised the BCG research finding that roughly 14% of AI-heavy knowledge workers reported symptom clusters around mental fog, slower decisions, and the need to step away from screens to reset. The Help Net Security coverage of the HBR-aligned analysis reported information overload around 19% higher and mental fatigue around 12% higher in the most-AI-exposed cohort. Both are self-reported workplace survey numbers, not lab data — that limit matters — but the pattern is real and it is most pronounced in roles with constant AI evaluation responsibilities. Executive roles are at the top of that exposure profile.
The 2024 MDPI Societies study (MDPI) on AI and critical-thinking offloading is the older mechanism reference. The study found an inverse correlation between AI usage frequency and self-reported critical-thinking engagement on knowledge tasks. Survey correlations cannot prove causation; the direction is consistent with the older cognitive-offloading literature.
The 2025 Microsoft Research / CHI study by Lee, Sarkar and colleagues is the most-cited recent reference on AI mediation and critical thinking among knowledge workers. The published paper surveyed several hundred knowledge workers and found that higher self-reported confidence in personal skill was associated with more critical thinking on AI-mediated tasks, while higher confidence in the AI was associated with less. The honest summary — without inventing precise numbers — is that the relationship between AI use and critical thinking is moderated by who the user trusts. Executives who trust their own judgment more than the AI's do more thinking. That direction matters for how the role is performed.
The deliberate-practice literature (Ericsson, Krampe & Tesch-Römer, 1993) is the constructive frame: specific cognitive skills respond to daily, effortful, on-purpose engagement. Judgment is one of those skills.
The team-level modelling effect
The executive case has a piece the individual-contributor case does not. How you use AI is modelled by everyone reporting to you. If your visible practice is to ask AI for the answer and accept it, the team learns that the originating cognitive act is the AI's job. If your visible practice is to decide first and then pressure-test, the team learns that the originating act is theirs.
This is not a soft observation. The way leaders demonstrate cognitive engagement with hard decisions is one of the strongest signals a team has about what good work looks like in the new tooling. Most companies have not figured out an AI-use norm yet. The executives who are most visible about how they use AI — and how they don't — set the norm by default.
A daily practice for AI-saturated calendars
The routine below is built for a real executive's working day. None of it requires you to use AI less. All of it requires roughly twenty minutes a day, mostly inside decisions you would have been making anyway.
1. Decide before you ask. Write a one-sentence call on every non-trivial decision before opening AI. Even if you change the call after the AI input, the originating act has happened. The pattern Microsoft Research observed — confidence in your own skill predicting more critical thinking — is built by exactly this habit.
2. Articulate before reading. When you ask AI for analysis, write your own three-bullet version of the reasoning first. Then compare. The comparison is the rep; reading the AI version cold is not.
3. Protect one no-AI thinking block a day. Twenty-five minutes, on the calendar, with no AI open. Use it for the hardest pending decision. The block does not need to produce an artefact. The unmediated thinking has to exist somewhere on the day.
4. Be visible about how you use AI. Tell the team what you used AI for and what you did yourself. Not as policing — as modelling. The cognitive norm of the team is set by what leadership is observably doing.
5. Treat the recovery side seriously. The 2026 workplace fatigue data is about evaluation load. Mid-afternoon screen-off windows, end-of-day boundaries, and protected non-AI working time are not soft benefits. They are the operating discipline that keeps the judgment muscle warm over months.
What this is not
A few honest disclaimers, because the executive AI category is loud.
This is not a claim that AI mediation makes executives less capable. The published evidence does not support that claim. The narrower claim — that the originating act of judgment, like any cognitive act, weakens when it gets fewer reps — is what the cognitive-offloading literature supports.
It does not say you should use less AI. The productivity gains are real and there is no defensible competitive posture in 2026 that ignores them. The recommendation is calibration, not abstinence.
It does not promise that this routine prevents AI brain fry or AI fatigue. The 2026 workplace data is about a pattern of cumulative evaluation load. The routine reduces some of the evaluation load by re-introducing originating thinking, but the workplace pattern is bigger than what an individual practice can fix.
How Senwitt fits
Senwitt is a daily seven-minute Set across six Senwitt Skills. For executives, the Reasoning Skill and the Writing Skill carry most of the load, but the mix is the point — judgment is built and maintained across multiple cognitive surfaces, not a single one. The deliberate-practice frame Senwitt is built on is taken from the Ericsson 1993 paper.
The for executives page and the for knowledge workers page lay out the case in more detail. The research/ai-overreliance page covers the workplace-level pattern.
One last observation that matters disproportionately for senior roles. The published workplace data on AI use is largely about individual contributors — engineers, analysts, designers, knowledge workers in the trenches. The data on executives specifically is thinner, partly because the cohort is smaller and partly because executives self-report less candidly when the questions touch their own cognitive performance. The honest read is that the executive case is the place in the org where the cognitive-offloading risk is structurally hardest to measure and structurally easiest to ignore. That is not a reason to panic. It is a reason for the calibration discipline to be self-imposed rather than externally measured: a daily routine you defend because the role demands it, not because a dashboard somewhere is flagging that your originating judgment has gotten thinner. The dashboard is unlikely to exist. The routine has to do the work of the dashboard.
