The AI Strategy Most Organizations Are Running Is the One That Got Them Stuck
Most organizations are not responding to AI. They are doing what they always do, and that is the problem.
A few weeks ago I went to the AWS Summit in London, and something I saw there has been sitting with me ever since.
The technology sessions were on huge stages with hundreds of seats. The People and Culture session had twenty seats and headphones, with about fifty of us standing around the edges trying to hear. It bothered me all the way home — not just because AWS got the staging wrong, but because the room was a picture of how most organizations are running their AI strategy right now. It looked exactly like every other response to every other wave of change I have watched executives navigate in the last twenty years.
Same playbook, new label
When I sit in executive rooms these days, I see roughly four patterns. Some organizations are putting lipstick on legacy, with twenty or thirty-year-old systems still running the business and AI bolted on top like a fresh coat of paint. Others are running toward AI the way they ran toward cloud — buy the platform, stand up the committee, announce the strategy, look busy. A third group is over-planning their way out of having to start, with endless vendor reviews and risk frameworks dressed up as governance.
And a fourth group is mostly listening to other executives doing the same three things. An echo chamber of the same conferences, consultants, and panels recycling the same playbook regardless of what is in front of it.
BCG surveyed a thousand executives last year and found three-quarters of companies have not yet figured out how to unlock value from AI. The playbook is not working, but everyone keeps running it anyway because it is the one they know how to run.
The AWS booth was a snapshot of all of it in one room.
The research has been on the table for over a year
Mercer just surveyed nearly twelve thousand executives, HR leaders, investors and employees worldwide for its Global Talent Trends 2026 report. The finding that should stop every executive cold:
“62% of employees say their leaders are underestimating the emotional and psychological impact of AI on their workforce. Only 19% of HR leaders are treating that as part of the AI plan.” - Mercer, Global Talent Trends 2026
That is a forty-three point gap between what the workforce is feeling and what HR is designing for. And it is not a knowledge gap. Mercer published these findings in January. BCG was making essentially the same argument eighteen months earlier: real AI transformation is two-thirds people and one-third technology. Most organizations are still running it the opposite way.
The research has been sitting on executives' desks for over a year, and the People and Culture booth at AWS was still that small. That tells you everything.
The question itself is the problem
Most of the executives I talk to are asking the wrong question. They are asking how do we implement AI, and that question is doing more damage than they recognize. It points the conversation at platforms, vendors, committees, governance frameworks, and adoption plans — all of which the existing organization already knows how to run. Which is exactly the problem.
The real question is a very different one.
What does this organization need to become to keep delivering what our customers need, in a world where AI is changing what is possible faster than our org chart can keep up?
That question points outward. It starts with the customer, not the platform. It assumes the organization itself must change, not just adopt another tool. The first question can be answered by a steering committee. The second one cannot.
The training response is the clearest example
Watch what most organizations are doing right now to address the people side of AI. They are buying more learning platforms, mandating more completions, and reporting completion rates to the board as if that constitutes readiness. It does not. That response was designed for a problem we do not have anymore.
The World Economic Forum's Future of Jobs Report 2025 asked employers worldwide which skills will matter most by 2030. The list reads nothing like what most learning platforms were built to deliver.
“Core skills for 2030: analytical thinking, creative thinking, resilience and flexibility, curiosity and lifelong learning, systems thinking, empathy, judgment, leadership and social influence.” - World Economic Forum, Future of Jobs Report 2025
None of these are training modules. You cannot mandate curiosity, install resilience through a video, or send someone to a workshop and have them walk out a systems thinker. These are capabilities, and capabilities like these are built differently. They need real enablement — coaching in the moment, practice on real work, leaders who model the behavior themselves and protect the conditions for it to grow. They develop when leaders change the conditions around the work: how decisions get made, how mistakes get treated, how questions get rewarded, and whether the team is being asked to think or just to deliver.
None of this means the capabilities are out of reach. They can be built. They just need a different approach than the one most organizations are still funding.
A different kind of change
This wave is not another version of cloud, or digital, or ERP. The shape, pace, and demands of it are all genuinely different from what came before, and you cannot run this one on the playbook that worked for the last one. You cannot answer it with the question that worked for the last one. And you certainly cannot build the workforce for it with the training infrastructure that was built for the last one.
Einstein said you cannot solve a problem with the same thinking that created it. He was not talking about AI, but he could have been.
Choose your hard
There are two paths from here.
The first is to keep running the old playbook with a new label. Buy the platform, announce the strategy, mandate the training, report the completion rates. Familiar, comfortable, easy to defend. A year from now, you will look very much like the organization next to you, and your customer will have a lot of options that look better than what you are offering.
The second is to stop and ask the harder question underneath it. What does this organization need to become to keep being worth choosing? That question is uncomfortable. It does not have a vendor demo, cannot be resolved in a steering committee, and is the only one that produces an answer different from the last twenty years.
Both paths are hard. One of them is hard now. The other is hard later, when the choice has already been made for you.