Berfin Yağız
3 min read
March 1, 2026
AI is no longer a technology companies are trying to adapt, it's a force every employee, every leader, and every organization is expected to work with. In 2026, AI isn't a future trend. It's the present. But the question isn't whether your company uses AI. It's whether your people actually know how to.
Let's take a step back before we look forward.
2025 might be one of the biggest years for technological change since, well, the internet. We saw entire industries begin shifting their workflows around AI tools, from marketing to finance, healthcare to education. Companies deployed AI copilots, built internal GPTs, and invested billions in infrastructure.
But here's the uncomfortable truth: most of that adoption stayed at the surface. According to McKinsey's 2025 Global Survey on AI, while 78% of companies reported using AI in at least one function, only 1 in 4 said their workforce was confident using these tools effectively. Gartner estimated that nearly 60% of AI initiatives stalled due to lack of user readiness, not tech readiness.
That mismatch, between ambition and ability, is the defining problem of AI in the enterprise. And it didn't get solved in 2025. It got louder.
If 2025 was about adoption, 2026 is about capability. The gap between "we use AI" and "our people understand AI" is now a measurable business risk. And it's starting to show up in the numbers.
PwC's 2025 CEO Survey found that 52% of CEOs see skill shortages as the number-one threat to growth, up from 40% the year before, with AI-related skills at the top of the list. LinkedIn's AI at Work Report confirmed that demand for AI literacy roles grew 3x faster than supply in the past 12 months. And Deloitte's Human Capital Trends report warned that organizations without structured AI upskilling programs are 2.4x more likely to face employee resistance during transformation.
In other words: AI is here. But human capability hasn't kept up.
The diagnosis is not the hard part anymore, so it is worth being blunt about what follows from it. If capability is the bottleneck, the work is not another tool rollout. It is four unglamorous things.
Find out where you actually stand. Not a survey asking people how confident they feel, but an assessment with enough headroom to show movement, taken before you start so you have something to compare against.
Train by role, not by seniority. Someone who reads dashboards, someone who builds them and someone who decides where AI gets used need different things. A single all-staff module serves none of them well.
Put the effort into engagement, not production value. Completion is decided by whether people have a reason to come back next week — a cohort, a challenge, a peer group, someone noticing when they stall — not by how polished the video is.
Use your own data and your own cases. The fastest way to lose a room is a generic example. The fastest way to keep it is a case they recognize from last Tuesday.
None of this needs a multi-year plan. Most of the programs we run start with one cohort, one business area and a baseline measurement — enough to prove the approach before scaling it. If you want to talk through what that would look like in your organization, get in touch.
On 24 September we are hosting the 10th Data & AI Literacy Round Table in Utrecht: HR and data leaders in one room, on who owns AI literacy and what real adoption takes. Invite-only, 30 to 40 seats.
Member of
Nominated
