Most AI rollouts still don't pay off, and the reason usually isn't the model
New reporting on why enterprise AI projects fail puts a number on it. A Gartner survey found only 28% deliver the return they promised, and one in five fail outright from being too ambitious or too loosely scoped for a machine to execute. A separate survey of 2,000 CIOs and CTOs found seven in ten say their teams are deploying AI faster than IT can track it. Among the failure patterns named, one stood out: teams that use the same model, or the same vendor's model, to both generate an answer and check it miss the errors a second, independent model would catch.
What it actually means
The number that matters here isn't 28%, it's what sits under it. Most of that failing fifth weren't beaten by a bad model. They were badly scoped from the start, pointed at a job nobody defined clearly enough for a machine to do. The fix that stood out in the research is one most teams skip: check a model's work with a different provider's model, because a vendor grading its own homework hides exactly the errors you need to catch.
If you run AI inside a business, that's the change worth making this week. Stop letting one provider mark its own work, and write down what the tool should and shouldn't do before it runs. Neither needs a new tool or a bigger budget, just a habit most teams haven't built yet.
A vendor grading its own homework hides exactly the errors you need to catch.
28%share of AI infrastructure projects that deliver the return they promised
Source: Gartner, via ABC17News (Stacker) · Sunday 27 September 2026