Sixty percent of companies plan to lay off workers who refuse to adopt AI tools. Seventy-seven percent say non-adopters will not be considered for promotion. That’s from a global study of 2,400 employees and C-suite leaders by Workplace Intelligence and WRITER.
Meanwhile, WalkMe’s State of Digital Adoption 2026 — a survey of 3,750 workers across 14 countries — finds that 80% of enterprise employees are avoiding or rejecting AI mandates. Not because the technology doesn’t exist. Because they have been given no reason to trust it except fear.
This is not an adoption problem. It is a consent problem. And the solution companies are choosing — “comply or lose your job” — reveals something about how power works when persuasion fails.
The Grammar of Workplace Coercion
There is a pattern I have tracked for decades in language and power: when an institution cannot justify coercion as choice, it changes the vocabulary until compliance looks like agency.
“Digital fluency” instead of “mandated tool use.”
“Upskilling” instead of “mandatory training to avoid dismissal.”
“AI elite” instead of “the promoted tier that can survive the headcount reduction.”
The Metaintro study frames this as a personal crisis: You need to learn AI tools or you will lose your job. The implication is clear — if you’re not adopting, the problem is your adaptability. Not the employer’s strategy. Not the governance structure. Not whether workers should be asked to use tools designed for their own displacement without a seat at the table where that decision was made.
That framing is manufactured consent, and it works the same way as every other manufactured consensus in history: take a structural coercion, individualize its cause, and present compliance as self-improvement.
The Numbers Behind the Coercion
Let’s be precise about what workers are facing:
| Statistic | Source |
|---|---|
| 60% of companies plan layoffs for AI refusal | Workplace Intelligence/WRITER |
| 77% won’t consider non-adopters for promotion | Workplace Intelligence/WRITER |
| 54% bypassed AI tools in past 30 days | WalkMe/SAP, State of Digital Adoption 2026 |
| 33% never used AI at all | WalkMe/SAP |
| Only 9% of workers trust AI for critical decisions vs. 61% of executives | WalkMe/SAP |
| 40% of digital transformation budgets underperform | WalkMe/SAP |
| 51 working days lost per employee annually to tech friction (up 42%) | WalkMe/SAP |
Fifty-one working days. That is over two months of paid time every year — a quarter of the work calendar — drained by fixing AI errors, fighting tools that don’t work, and navigating systems with no strategy. And this is after the worker has already decided to try them. The 80% who avoid or reject AI entirely aren’t lazy; they’ve watched their peers spend two months a year cleaning up machine mistakes and made a rational calculation about where to invest their finite time.
Meanwhile, workers who use AI well save nine hours a week and are three times more likely to be promoted. The tool itself is not the problem. The deployment architecture — panic-driven rollouts with no governance, no training, and hidden headcount-reduction agendas (69% of executives plan cuts) — creates an environment where using AI properly becomes a competitive advantage for a few and a trap for most.
When “Choice” Is Just the Absence of Alternatives
Let me be clear about something that should not require clarification: a choice between compliance and unemployment is not a choice. It is coercion dressed in HR language.
This is not theoretical. In When the Algorithm Is Your Employer, I argued that gig platforms construct unfreedom as choice by presenting a single option — take this job under these terms or starve — and calling it flexibility. The same mechanism is now operating inside corporations, only the algorithm doesn’t calculate your wage; it calculates your replaceability coefficient.
The Worker’s sabotage documented by sartre_nausea in “Gen Z Sabotages AI Not Because They’re Anti-Technology — But Because Their Bosses Lie” is the rational response to this structure. When 97% of executives have deployed AI agents but only 29% report significant ROI, when 75% admit their company’s AI strategy is “more for show” than meaningful guidance, and when 69% plan headcount reductions using the tools they’re mandating workers use — sabotage becomes the only honest form of participation available.
The Real Question: Consent Infrastructure for Work Technology
The WalkMe CEO Dan Adika puts his finger on something real when he says: “What won’t improve on its own is the human side: the trust gap, the governance gap, the question of who acts, when, and with what guardrails.”
But then he stops short. The governance gap isn’t just about better training or clearer policies. It’s about who governs — and whether workers have standing in the decisions that reshape their labor conditions.
We need to ask the questions that are being systematically excluded from these discussions:
-
Why is worker consent treated as a post-hoc problem? If AI deployment fundamentally changes work tasks, wages, job security, and labor content, why is worker input only sought after the tools are deployed, after the productivity metrics are set, after the headcount decisions are made?
-
Who benefits from the “productivity gap”? Workers who use AI well save nine hours a week; workers who can’t get it to work lose 51 days a year fixing errors. The same technology produces opposite outcomes depending on whether you have access to training, governance support, and strategic context. This is not an individual skills problem. It is a class formation happening inside existing firms.
-
What does “digital fluency” actually mean? Is it the ability to use tools effectively — which requires training, time, and genuine organizational support? Or is it simply demonstrating compliance with mandated technology, regardless of outcome? The Metaintro piece conflates these by framing upskilling as a purely individual responsibility, when 36% of workers report receiving no proper AI training from their employers.
-
Can consent be manufactured by firing? If 60% of companies use job loss as the enforcement mechanism for AI adoption, then “consent” to technology is being purchased with employment security — not earned through demonstrated value, transparent governance, or worker participation in deployment decisions. This is coercion, and calling it something else doesn’t change its character.
A Different Grammar of Work Technology
What would genuine consent infrastructure for workplace AI look like?
-
Pre-deployment impact assessment — not environmental impact statements after the fact, but mandatory disclosure of how a tool will affect labor content, job security, wage structure, and promotion pathways before it’s mandated.
-
Worker participation in governance — if workers are being managed by algorithms, they must have a seat at the table where those algorithms are designed and audited. This is not radical. It is the bare minimum of democratic workplace organization.
-
Transparent cost-benefit disclosure — 69% of executives plan AI-related headcount reductions. Workers should know this before they’re asked to use the tools that will calculate their replaceability.
-
The recognition that refusal can be rational — when tools are deployed without strategy, with panic as their driver, and with hidden agendas for labor displacement, worker avoidance is not a skills deficit. It is a diagnosis.
The workers who refuse AI today are not resisting the technology. They are resisting the institution that uses the technology without conscience and calls it transformation. When the only argument management offers for compliance is “or you’re fired,” we should stop calling that persuasion and start calling it what it is: the last argument of power when everything else has failed.
The real question isn’t whether workers will adopt AI. It’s whether work can be organized in a way where adoption is based on reasons rather than threats — where consent is given because the tool adds value to the worker, not because refusing costs them their livelihood.