AI Change Stalls Until Every Leader Gives the Same Answer on Jobs
Nov 07, 2026
|
Key Takeaways â–¶ Most AI programmes stall for a reason that never appears in the plan: people don’t know what AI means for their jobs, so they use it quietly and say nothing â–¶ Until every leader gives the same answer on jobs, the quiet use carries on, whatever you spend on licences and training â–¶ The answer has five parts: what AI is for here, what stays with people, what gets checked, what happens if a role changes, and what nobody knows yet â–¶ Managers get it a week before anyone else. Then every leader says it the same way and shows it in their own work |
Most AI programmes I see stall in the same place. The licences are bought, the training is booked, and six months on the work is done the way it always was. The usual explanation is that people need more training. In my experience the problem starts earlier. People don’t know what AI means for their jobs, so they’re using it and keeping quiet about it.
Here’s the test. Ask five of your leaders, separately, what they say when someone on their team asks whether AI puts their job at risk. If nobody has agreed an answer, each of them improvises one. Some reassure. Some say every job will change, so keep learning. Some say, honestly, that they don’t know. Each answer is fair on its own. Heard together across one organisation, they tell people that the top hasn’t decided. And people who think the top hasn’t decided use AI, take the time it saves, and say nothing.
Until every leader gives the same answer on jobs, that is what your people will do.
Why Do People Keep Their AI Use Quiet?
Because the question that matters most to them hasn’t been answered. Someone who gets good at AI doesn’t know whether that makes them more valuable or easier to replace. The share of employees worried about losing their job to AI has gone from 28% to 40% in two years (Mercer, 2026). Someone carrying that worry hears a different answer from each leader and draws one of two conclusions: nobody has decided, or somebody has decided and isn’t saying. Either way, keeping your head down looks sensible.
A course can’t fix this. A usage dashboard can’t see it, because it only counts the approved tool. The answer on jobs is a leadership decision, and it has to come before the rest of the plan. One answer, from every leader, said the same way, before anything else.
|
UPSKILL, the weekly newsletter What’s actually working with AI at work One email a week. Prompts, skills and what we’re seeing inside real businesses. |
What One Answer on Jobs Is Made Of
Leaders can’t promise that no job will change, and people can tell when a promise can’t be kept. What a leadership team can do is agree one honest answer and repeat it. In our work it has five parts, and the hardest one to write is the last.
Most leadership teams find the middle parts quick. Ask which decisions no tool should make here and the list comes fast: hiring, pay, performance, money. Ask what gets checked before anything goes out and everyone says a person reads it first. Those lines already exist in most leaders’ heads. Writing them down is an hour’s work, and very few teams have done it. Tap through what the full answer looks like.
|
Interactive: the one answer What one answer on jobs is made of Five parts. Tap each for example wording. The parts are G3NR8’s guidance from leading change with AI. The sentences are examples, to be rewritten in your own words. What it’s forWhat stays with peopleWhat gets checkedIf a role changesWhat we don’t know
Leave the headcount forecast out. Nobody can keep one, and people can tell when the honest sentence about what nobody knows is missing. Source: G3NR8 guidance; example wording is ours |
Managers Get It First
The answer reaches people through their manager, because that’s who gets asked, in a one to one, on a Tuesday. Managers get the question first and are usually the least prepared for it. So they get the answer before anyone else, in the agreed words, with a chance to practise saying it out loud and a route back for the questions they can’t answer.
The order matters. If a team hears about the AI plan in the same meeting as its manager, the manager has nothing to say when the questions start that afternoon. Hold the launch until the managers have the answer.
Leaders Show Their Own Use
Then leaders have to be seen living by it. Most senior leaders already use AI, and most of that use happens in the inbox and the board pack, where nobody else sees it. Their teams see a leader who talks about AI and never visibly uses it, and they draw their own conclusions. Gallup asked employees whether their manager supports AI use. Where people strongly agreed, 78% used AI a few times a week or more. Where they didn’t, 44% did (Gallup, 2026).
|
Where the manager backs it Frequent AI use, by whether the manager supports it Share of employees who use AI a few times a week or more, by whether they strongly agree their manager supports AI use.
Source: Gallup, AI and Workplace Productivity: What the Data Show, July 2026 (Q2 2026 data) |
The cheapest version is a meeting the leader already runs. Have AI pull the action list from the transcript of the last one, read it in front of the team, correct what it got wrong, then run the meeting from it. The team sees the tool used and sees the checking rule followed, every week, with no slide.
What Leaders Do This Month
|
1. Ask every leader the question, separately Ten minutes each, by voice or in writing: what do you tell someone who asks if AI puts their job at risk, what stays with people, what gets checked, what happens if a role changes. Separately, because in a room together leaders settle on wording vague enough to cover everyone. |
|
2. Write down what matched Usually most of it. One page, in plain words, plus the honest sentence about what nobody knows yet and how people will hear when it changes. Leave the forecast out. |
|
3. Managers get it a week before anyone else They get the question first, so they get the answer first, with a run-through of the conversation and a route back for the questions they can’t answer. |
|
4. Every leader says it, then shows it The same lines in every team meeting. Then one piece of the leader’s own work done with AI in front of the team and checked in front of the team. The meeting action list is a good first one. |
|
5. Ask again in three months The tools will have changed. Ask the same leaders the same question, count the answers, and add a line if the business has learned something. |
If you want to know where your organisation stands, ask five of your leaders this week. Count the answers. One, and the first job is done. More than one, and it goes to the top of the leadership agenda.
How many answers would you get?
|
Free report: The AI Adoption Gap 2026 Why AI use is everywhere and results are flat, what the leading organisations do differently, and the questions to ask your leadership team. Written by people who have trained 150+ large organisations. |
Sources & References
| Mercer: Global Talent Trends 2026, 25 February 2026. Employee concern about job loss due to AI rose from 28% in 2024 to 40% in 2026. mercer.com |
| Gallup: AI and Workplace Productivity: What the Data Show, 27 July 2026, updated 30 September 2026, Q2 2026 data. 78% of employees use AI frequently where they strongly agree their manager supports AI use; 44% where they do not. gallup.com |
Frequently Asked Questions
|
What should leaders tell employees about AI and jobs? One agreed answer, said the same way by every leader: what AI is for in the business, what stays with people (hiring, pay, performance and money), what gets checked before anything goes out, what happens if a role changes, and an honest line about what nobody knows yet. Headcount forecasts are left out, because nobody can keep them. |
|
Why do employees hide their AI use at work? Because they don’t know whether being good at AI makes them more valuable or easier to replace. Worry about losing a job to AI has risen from 28% to 40% since 2024 (Mercer, 2026). Until every leader gives the same answer on jobs, keeping quiet is the safe option. |
|
How do you get a leadership team to agree one message on AI? Ask each leader separately what they tell people, then compare the answers. Most leadership teams already agree on what stays with people and what gets checked, and have never written it down. Write down what matched, add the honest line about what nobody knows, and have every leader say the same thing. |
|
What is the line manager’s role in AI change? Managers get the job question first, in one to ones, so they need the answer before their teams do and a chance to practise it. Gallup’s 2026 data shows 78% of employees use AI frequently where their manager strongly supports it, against 44% where the manager doesn’t, so what the manager visibly does has a large effect on whether the team uses AI. |
|
Does traditional change management still work for AI? The basics hold: a sponsor, one clear message, managers equipped, reinforcement and measurement. What’s different with AI is the order. People started using it before the organisation did, so the first job is the answer on jobs, before the training and the tools. |
Roy Murphy is co-founder of G3NR8, which has trained 150+ large organisations to use AI on their real work. This post sets out G3NR8’s guidance on leading change with AI, alongside the published research listed above. Published 6 November 2026.
Get UPSKILL, our weekly AI newsletter.
What’s actually working with AI at work. One email a week: prompts, skills and what we’re seeing inside real businesses.
