The AI Urgency Trap: Why 95% of AI Projects Fail
17 lug 2026
I have been repeating this in every room. HBR just put numbers on it.
Lots of companies are not really running an AI strategy. They are just switching tools on.
I mean, they are not choosing AI. They are reacting to it.
Fear of falling behind. Pressure to look up-to-date. Trying to keep up with what they read about themselves in the press.
David De Cremer wrote about it in Harvard Business Review this month. He calls it the urgency trap. And the numbers are hard to look away from.
→ 95% of generative AI projects fail (MIT)
→ More than 6,000 senior executives across four countries. Roughly 90% saw no measurable productivity gain from AI in three years.
Three years. Almost nothing to show.
Here is what I think is really going on.
Most companies reach for AI to cut costs, and to do that now. That’s a valid claim. But protects the status quo. It does not grow anything.
And most of these projects sit with the tech team. So the attention goes to the tools, not to the people who actually have to use them.
The real edge was never speed. It is clarity about why you exist and what you want to achieve, for your customers and for the people who work with you.
Before you scale the next AI initiative, three questions worth sitting with:
→ Are you moving fast because it solves a real problem, or because moving fast feels safer?
→ What skills, habits and culture need to be in place first?
→ Who needs time to learn and trust this, and are you giving it to them?
Moving first is easy. Moving with purpose is the hard part.
Are you solving a real problem, or just keeping up?
Why the tech team cannot fix a people problem alone
The pattern De Cremer names in HBR is the same one I see up close in my Private AI Advisory work. Company after company hands AI to IT or a small innovation team, measures success by how many tools got deployed, and wonders eighteen months later why nothing changed in how the business actually runs.
The fix is not a better rollout plan.
→ The organisations that get real value treat AI adoption as a leadership and culture question first, and a tooling question second.
→ They spend real time on what skills, habits, and trust need to exist before the tool lands, not after. That is slower at the start and faster everywhere after.
I wrote about this exact pressure, the feeling that you have to move now or fall behind, in from AI pressure to AI confidence. The short version: the leaders who get past the urgency trap are the ones who let themselves slow down long enough to ask why, before they ask how fast.
If you want a structured way to sit with those three questions yourself before your next AI decision, I have a free AI Coach Tünde Light, a set of self-coaching prompts built for exactly this kind of reflection. Leave your email here and I will send it over: https://www.thechangerepublic.com/free-resources