Everyone's Using AI. Nothing's Changed.
Oct 08, 2026
|
Key Takeaways â–¶ 80% of people using AI at work say it’s made them more productive. Only 37% of organisations see any profit from it, the same as last year (McKinsey, 2026) â–¶ The time AI saves goes back to the person who saved it and gets spent there. The report that was due Friday is still due Friday â–¶ Only 13% of AI users think they’d be rewarded for changing how they work if short-term results slipped, so most keep quiet (Microsoft, 2026) â–¶ The organisations making money from AI changed what they ask their teams to deliver. Three questions tell you if yours has |
|
80% of people who use AI at work say it’s made them more productive (McKinsey, 2026) |
37% of organisations see any profit from AI, the same as 2025 (McKinsey, 2026) |
13% of AI users think they’d be rewarded for changing how they work if results slipped (Microsoft, 2026) |
Half of American employees now use AI at work. Eight in ten of the people using it say it’s made them more productive.
So how many organisations can point to any profit from it? 37%. Same as a year ago.
Three years of licences, rollouts and lunchtime webinars, and that number hasn’t moved.
People are using it. It works for them. Nobody’s stopping them. So where’s the gain going?
Straight back to the person who made it. And it stays with them.
That’s what this series is about. Six posts on the gap between people using AI and organisations getting anything back from it. This one covers how the gap happens. The next five go role by role: leaders, managers, HR and L&D, operations, and the AI use nobody can see.
Where the Three Hours Go
Take one task. A monthly report that used to take four hours now takes one, because she drafts it with AI and checks it. Ask her in a survey and she’ll tell you AI has made her more productive. She’s right.
Now look at the same month from three desks.
|
The person writing the report Saves three hours. Spends them on a bit more polish, a quieter afternoon and the next job on the list. Nobody asked what she’d do with the time, so she decided for herself. |
|
Her manager Gets the same report on the same Friday, from the same team, against the same targets. Has no idea it now takes an hour. |
|
The finance director Sees the same headcount, the same cost line and the same output. Signs off another year of AI licences and wonders where the return went. |
Nobody did anything wrong. The time was saved, then quietly used up, because nothing the organisation asked for had changed.
BCG put numbers on it in June. Of frontline staff who use AI regularly, 42% save at least eight hours a week. A full working day. And 66% get little or no guidance on what to do with it.
|
66% of frontline staff who use AI regularly get little or no guidance on what to do with the time it saves. 42% of them save at least a full working day a week. Source: BCG, AI at Work 2026: Strategy Matters More Than Tools (11,749 workers, 14 markets, June 2026) |
A spare day a week and nobody’s said what it’s for. Where do you think it goes?
An organisation only sees the gain when it changes what it asks for. A faster turnaround. A new deliverable. A smaller team handling the same volume. Until one of those moves, the gain belongs to the employee and never reaches the accounts. Buying more licences won’t change that.
Why Nobody Owns Up
So why don’t people just say it? “This takes me an hour now. What do you want me to do with the other three?”
Put yourself in their chair. Say that to your manager and one of three things happens. You get more work. Your job looks smaller. Or someone starts checking your output a lot more closely.
Keep quiet and none of that happens.
Microsoft asked 20,000 people who already use AI at work about exactly this in May. Only 13% said they’d be rewarded for changing how they work with AI if short-term results slipped. The other 87% are doing the sensible thing.
|
13% of people using AI at work think they’d be rewarded for changing how they work with it if short-term results slipped. Source: Microsoft, 2026 Work Trend Index Annual Report (20,000 AI users, 10 markets including the UK, May 2026) |
We see it in teams before we’ve run a minute of training.
|
G3NR8 data: one team, before training Earlier this year we surveyed a creative team of 11 at a major UK retailer before any training. All 11 were already using AI for work. Seven called themselves the “AI Underground”: good at it, and keeping it quiet. Nobody rated their own understanding above 3 out of 5. All 11 asked for training first, and seven asked for a clear company view on what AI is for. |
Everyone using it. Hardly any of it visible. And the people running the team had no idea how much was going on.
What the 6% Did Differently
McKinsey picks out a small group it calls high performers: organisations getting 5% or more of their operating profit from AI. About 6% of companies. Same as last year.
What sets them apart? Nearly three-quarters of them have fundamentally redesigned how work gets done because of AI. Among everyone else it’s about a quarter.
“Redesigned” sounds grand. In practice it looks like this.
|
The monthly report Now written up as a one-hour job, and the deadline moved from Friday to Tuesday. |
|
The quote queue Was a team of six, now four. The other two moved onto the account work nobody had time for. |
|
The customer query Turnaround cut from two days to one, and the service standard rewritten to say so. |
In each case the organisation now expects something different. So the saved time has somewhere to go, and someone to count it.
I spend my weeks putting AI into large organisations and I see the same thing every time. The teams that get somewhere have someone senior who changed what the team delivers, and told them out loud.
The Collection Test
If you lead a team, this is the useful bit. Three questions. Ask them of any team this week. Ten minutes.
|
1. Name the task Which piece of recurring work is done differently now because of AI? Ask for the task, what changed and who signed it off. If you get a list of tools back, nothing’s changed yet. |
|
2. Where did the hours go? If that task dropped from four hours to one, who decided what the other three are for? If nobody did, they’ve gone. Decide up front: more customer time, a faster turnaround, a shorter queue. Pick one and write it down. |
|
3. What changed in what you ask for? Has a target, a deadline, a service standard, a team shape or a deliverable changed because of AI? That’s the only place the gain shows up. If nothing you ask for has changed, nothing you measure will either. |
Run it on three teams and you’ll find your gap. In most organisations the honest answer to question three is “nothing”, and that one answer explains the 37%.
Your people have already done their half, on their own and mostly quietly. The other half is a handful of decisions about what you ask them to deliver. Those are yours.
Which task in your organisation is specified differently today than it was a year ago, because of AI? If you can’t name one, where do you think the time is going?
Next in the series: leaders. They’re now the heaviest AI users in the building, and their teams still can’t tell what they want. Next week, why one doesn’t fix the other.
|
FREE REPORT: The AI Adoption Gap 2026 Everyone’s using AI. So why isn’t anything changing? We’ve pulled the whole series into one report you can take into your next leadership meeting. âś… The 2026 numbers from Gallup, McKinsey, Microsoft and BCG, all sourced âś… What the 6% making money from AI did differently âś… Leaders, managers, HR and operations: where each one is stuck âś… The collection test, ready to run with your teams this week G3NR8 has trained 150+ large organisations. This is what we see in every one of them. |
|
Get BAD AI in your inbox every week Our free weekly newsletter on what AI’s actually doing at work. The bad, the useful and the tools worth trying. Five minutes, Thursday lunchtime. |
Sources & References
| Gallup: Global Indicator: Artificial Intelligence, data as of May 2026 - 52% of US employees use AI in their role at least a few times a year (21% in 2023). gallup.com |
| McKinsey & Company: The state of AI in 2026: On the road to ROI, August 2026 (1,719 respondents, 97 countries, fielded 4 May to 8 June 2026) - 37% attribute at least some EBIT impact to AI, about the same as 2025; 6% high performers (5%+ of EBIT), flat; 80% of AI users report improved individual productivity; nearly three-quarters of high performers have fundamentally redesigned workflows, against about a quarter of others. Figures as reported by The Register, 25 August 2026. theregister.com |
| Microsoft: 2026 Work Trend Index Annual Report: Agents, human agency, and the opportunity for every organization, 5 May 2026 (20,000 AI-using knowledge workers, 10 markets including the UK) - 13% rewarded for reinventing their work with AI if short-term results aren’t met. microsoft.com |
| BCG: AI at Work 2026: Strategy Matters More Than Tools, 3 June 2026 (11,749 workers, 14 markets) - 42% of frontline regular AI users save at least 8 hours a week; 66% receive limited or no guidance on what to do with the time saved. bcg.com |
| G3NR8: Pre-training survey, creative team of 11 at a major UK retailer, 2026 - 11 of 11 using AI for work; 7 of 11 self-described “AI Underground”; self-rated understanding average 2.7 out of 5, none above 3; 11 of 11 named training as their first ask, 7 of 11 a clear company AI vision. |
Frequently Asked Questions
|
What is the AI adoption gap? It’s the gap between what people already do with AI at work and what their organisation has changed in what it asks of them. Gallup says 52% of US employees used AI in their role as of May 2026, and in McKinsey’s 2026 State of AI survey 80% of AI users said it made them more productive. Yet only 37% of organisations put any profit impact down to AI, about the same as 2025. People are using it. The organisation hasn’t collected the gain. |
|
Why does AI make people more productive without improving company results? Because the time saved goes back to the person who saved it, and gets spent there. If a four-hour task now takes one, but the deadline, the deliverable, the team size and the targets haven’t moved, the organisation gets exactly the same output. The three hours go on more polish, an earlier finish or the next job. BCG’s AI at Work 2026 survey found 42% of frontline staff who use AI regularly save at least eight hours a week, and 66% get little or no guidance on what to do with that time. |
|
Why don’t employees tell their managers how much time AI saves them? Because it rarely pays to. Microsoft’s 2026 Work Trend Index, based on 20,000 people who use AI at work across 10 markets including the UK, found only 13% think they’d be rewarded for changing how they work with AI if short-term results slipped. Owning up to a saving usually means more work, a smaller-looking job or closer checks on your output. Keeping quiet and hitting the usual deadline carries none of that risk, so most people keep quiet. |
|
What do the organisations making money from AI do differently? They change what they ask their teams to deliver. McKinsey’s 2026 high performers, the roughly 6% of organisations getting 5% or more of their operating profit from AI, stand out for redesigning how work gets done: nearly three-quarters say they’ve fundamentally redesigned workflows because of AI, against about a quarter of everyone else. In practice that means a deadline, a deliverable, a service standard or a team shape has changed, so the saved time has somewhere to go. |
|
How can a leader tell whether AI is actually changing their organisation? Ask any team three questions. Which specific recurring task is done differently now because of AI? Where did the saved hours go, and who decided what they were for? What has changed in what the organisation asks for: a target, a deadline, a service standard, a team shape or a deliverable? If nothing you ask for has changed, nothing you measure will either, however much AI your people use. |
This is Part 1 of 6 in The Adoption Gap series, on what the 2026 research says about who’s using AI at work and what it’s actually changing. Published 8 October 2026.
Keep informed with the newsletter for PE operating partners and the portfolio companies they back.
Â
Get operational insights and trends, AI frameworks, resources and real deployment stories.
