Organizations do not resist flexibility because they prefer offices. They resist because flexible work takes away the one instrument of supervision many managers had — watching people work — and puts nothing in its place. That is why the argument was never settled by data: what is at stake is not evidence, it is what is left for the manager to do.
Where resistance to remote work and to AI converge
Opposition to remote work does not come from a single place, and treating it as one bloc stalls the project. There are four profiles, and the same person may carry traits of more than one.
- they know the benefits and challenges of remote work in detail and oppose it anyway, out of personal interest in the current arrangement — remote work keeps the source of the money and decentralizes where it lands, which mobilizes anyone who feels threatened by the loss, from local businesses to commercial real estate. They act subtly, undermining the rollout rather than contesting it. This is not resolved by technical debate, but by transparency: justify decisions in public and keep conflicts of interest from contaminating the choice.
- Resistant for personal or cultural reasons
- they prefer the office routine and direct, visible supervision, and believe physical presence is essential to productivity. They have no hidden agenda — they have habit. They respond to awareness-building, training and periodic in-person gatherings.
- Generational and technological
- they face genuine difficulty with the tools and unease with the new dynamic. The answer is digital inclusion, gradual adaptation and technical support — not persuasion.
- Underinformed or misinformed
- they form an opinion from limited experience, outdated information or hearsay. This is the only group where communication and education actually solve the problem.
And why do these same profiles also resist AI? The manager who does not trust what he cannot see is the same one; the one who fears losing his function is the same, now with a more concrete reason. And the instrument that solves it is also the same: with no way to watch the work happening, you need a plan, a deadline, a recorded commitment and an assessment at the close of the cycle. Anyone who has already built that for distributed people has nothing to invent in order to supervise an AI — all that is left is plugging the new participant into the existing artifacts.
The table below is not research: it is a mapping to test against your own organization.
| Profile | On remote work | On AI |
|---|---|---|
| Well informed with a hidden agenda | "We need people together" | "It is not the right moment yet" — with their position anchored in today's operations |
| Personal or cultural | Physical presence is essential to productivity | If I do not see someone doing it, I do not trust what was done |
| Generational and technological | Genuine difficulty with the tools | Same difficulty, new tool |
| Underinformed or misinformed | Opinion formed by hearsay | Opinion formed by the latest headline in the feed |
Where the analogy ends
Pushing the parallel too far would be dishonest. A remote worker does not start making up data because they are at home. A language model does: it errs with confidence, does not know when it does not know, and can be swayed by content it read itself in the middle of the task. Distrust of AI has a basis that distrust of remote work never had.
What repeats is the shape of the resistance, not the legitimacy of each objection. And the practical conclusion is the opposite of "trust and delegate": AI work demands explicit, continuous supervision with a human owner — more supervision than an experienced professional gets, not less. The difference is that there is now a way to exercise it that is not looking across the office.
The physical office is not a productivity mechanism
Remote work showed that tasks can be completed in less time. The tension appeared when organizations realized the worker had the autonomy to capture that gain for themselves — as flexibility, or as fewer hours.
Confining people in offices does not raise productivity. Often the effect is the opposite: the same workload stretched over longer hours, purely to fill the conventional schedule. Managers and their reports find creative ways to fill the day with activities that look like work and produce no result. And when the manager shares the same space as the team, there is a tendency to equate presence with productivity, neglecting management by results.
The office is not a productivity mechanism: it is a visibility mechanism, comfortable enough to postpone building anything else. The alternative is managing teams around deliverables, with the deliverables plan filling the gap between strategy and operational tasks. Without a structure that puts outcome above hours, the gain comes back as pressure.
And AI takes that argument to its limit. An AI worker has no office to show up at, no working hours to put in, no look of concentration to display. The visibility mechanism was not taken away from it — it never existed. Anyone who held management together by looking at the office is not left without one tool among several: they are left without the only one they had, and discover that the substitute they postponed for years is exactly what they now have to build.
What distance demanded of leadership
Remote work is not a logistical adjustment; it is a holistic change, and the bill arrives first for whoever leads. The central shift is moving away from physical presence and toward results and progressive deliverables — which sounds simple and is not, because an adaptive challenge cannot be solved by a technical solution alone. Buying videoconferencing is technical. Relearning how to tell whether the work is moving is adaptive.
- Alternate between strategic contemplation and hands-on engagement. Leadership in this phase looks like improvisational art: guided by clear values, marked by unpredictable moves.
- Recruit partners and allies, instead of driving the change in isolation — a network that brings diverse readings and protects the initiative.
- Acknowledge losses. People naturally resist the new; leaders need to empathize publicly and lead by example.
- Keep self-awareness, separating the professional from the personal and handling criticism strategically.
AI adoption is being handled in the same wrong way: as a purchase. Choosing the model, buying the licenses and switching on access is the technical part, and it is the part everyone does first. The adaptive part is the same one from 2020 — relearning how to know whether the work is moving without seeing anyone work — with one aggravating factor: the four points above apply without translation. "Acknowledging losses" stops being generic empathy when part of what is lost is the mechanical work that used to fill the manager's own day.
None of this is confined to remote work: results-centered management sharpens the focus on collective goals under any arrangement — remote work merely catalyzes practices that were already loose.
The rehearsal and the premiere
Where resistance to flexible work was resolved, organizations stopped arguing about the place and started writing down what had been assumed: what the team committed to deliver, who answers for what, by when, and how it was assessed. As long as there was an office to look at, you could get away with not writing it.
Whoever did that has the premiere won. Whoever simply waited for people to come back to the office will find the same bill on the desk — now with no option to wait. The remote worker could be called back. AI has nowhere to be called back to.
The test, then, is not what you think of the technology. It is whether there is, in your organization, any way of knowing the work was done that does not depend on watching someone do it.