AI is making people faster. It is not yet making most organisations better.
That is the uncomfortable headline from McKinsey’s latest State of AI survey, which gathered responses from 1,719 people across 97 countries in mid-2026.
Eight in ten respondents say AI has improved their personal productivity. Half say it helps them make better decisions.
Yet the share of organisations reporting any contribution to profit from AI has barely moved in a year. It sits at 37%.
More people using AI. More money going in. Roughly the same results coming out.
That gap is not a technology problem. It is a leadership one.
Why This Matters Now
The investment curve is steep, and it is not slowing:
- 44% of organisations now say AI is scaling across the enterprise, up from 38% last year
- 60% expect to increase AI investment in the next twelve months
- Nearly a third have chosen to build software in-house with AI rather than buy it
But the group McKinsey calls “AI high performers”, those seeing significant, measurable financial value, remains stuck at just 6% of respondents. Unchanged from 2025.
The researchers are clear about the reason. The technology is moving faster than organisations can absorb it. The bottleneck is no longer the tool. It is the people, habits and structures around it.
What the High Performers Do Differently
The 6% are not simply spending more. They are leading differently.
- They redesign work, not just add tools. Nearly three-quarters have fundamentally reshaped workflows around AI. Only a quarter of everyone else has.
- Their leaders go first. Senior leaders in high-performing organisations are twice as likely to visibly own and champion AI initiatives.
- They measure impact. They are twice as likely to have clear processes for tracking what AI actually delivers.
- They aim higher than efficiency. Almost everyone uses AI to cut costs. High performers also use it to grow and innovate: 74% pursue growth goals, compared with 47% of others.
- They decide where humans stay involved. They are deliberate about when a person must check, judge or approve what AI produces.
None of these are technical capabilities. Every one of them is a leadership behaviour.
Where Leaders Get Stuck
Collecting pilots instead of changing work. A new tool here, a trial there. Lots of activity, little transformation.
Squeezing the middle. This is the finding we think deserves far more attention. Among middle managers and individual contributors, 47% report at least one negative effect from AI at work, compared with 31% of executives. One in five middle managers say it hurts their ability to think critically. Nearly as many say it increases anxiety about their careers or piles on pressure to do more. The people expected to make AI work are the ones feeling its strain most.
Leading through fear. Last year, 32% of respondents expected AI to shrink their workforce. Only 14% actually saw that happen. Yet this year, 39% expect reductions. Uncertainty is growing faster than the reality, and unaddressed uncertainty quietly erodes trust.
Handing AI to IT. When AI is treated as a technology project, it gets managed like one. When it is treated as a change in how people work, it gets led.
Toolkit for Action
- Pick one workflow and rethink it from scratch. Do not ask “where can AI help?” Ask “if we designed this today, knowing what AI can do, what would it look like?”
- Go first, visibly. Use the tools yourself. Share what worked and what did not. Permission spreads from the top.
- Protect your middle managers. Check in on workload, focus and confidence. Give them space to learn, not just targets to hit.
- Measure outcomes, not usage. Logins and prompts are not impact. Track time saved, quality improved and value created.
- Name where the human decides. Agree, explicitly, which outputs need human judgement before they are acted on.
- Talk honestly about roles. Silence breeds speculation. Share what you know, what you do not, and how people will be supported.
5 Reflection Questions
- Where is AI making individuals faster in our organisation, but not making the organisation better?
- Have I personally modelled the way I want my team to use AI?
- How are our middle managers really experiencing this change?
- Are we measuring AI activity, or AI impact?
- What conversation about the future of roles are we avoiding?
How Mind-Gap Can Help
AI adoption succeeds or stalls on the quality of leadership around it. At Mind-Gap, we help leaders and managers build the confidence, clarity and change capability to turn new tools into genuine impact, through ILM-accredited leadership development, executive coaching and team programmes designed around your real challenges.
If your organisation is investing in AI but waiting for the results, let’s talk.
Get in touch here.
Original Research
McKinsey & Company, The State of AI in 2026: On the Road to ROI (August 2026)
Â
