10/12/2026
McKinsey’s latest AI research points to a gap that sounds very familiar to me: 80% of respondents say AI has improved their individual productivity, but only 37% report a positive EBIT impact and just 6% qualify as AI high performers.
I’ve seen versions of that gap for years in leadership development. People can learn something new, understand it and even use it well, but that still doesn’t mean the company will get the performance improvement it expected. What really caught my attention was that nearly three-quarters of McKinsey’s AI high performers had substantially redesigned workflows, compared with only about one-quarter of the other organizations.
That suggests the harder question isn’t simply whether people are using AI. It’s whether the work itself is changing. Are people applying AI to the work that matters? Have workflows or decision rights changed? What are managers reinforcing? What old work has actually stopped? And what business result should improve if all of this is working?
That is also why I think applied AI development has to go beyond teaching people how to use the tools. People need an opportunity to use AI against their actual responsibilities, workflows, decisions and business challenges, with enough reinforcement and feedback for the new way of working to stick.
Training can create capability. The real value comes when that capability becomes consistent ex*****on and measurable company performance.