The number every AI project should know
Current studies paint a remarkably consistent picture: depending on the survey, between 62 and 80% of AI projects in companies fail – not because the deployed technology doesn't work, but because the workforce doesn't adopt it, works around it, or actively rejects it. A Bitkom study puts the share of projects that failed due to workforce resistance at 62%; other surveys arrive at even higher figures once missing strategic preparation and culture change are factored in.
Why this gets overlooked so often
AI projects are usually budgeted and planned as a technical undertaking: which model, which vendor, which integration. The question of whether and how the people who'll work with it daily actually embrace this change rarely appears as its own line item in many project plans – it's silently assumed. That exact gap is why a technically flawless system still fails.
Failure rarely looks like open rejection
In practice, failure usually looks unspectacular: a tool that's introduced but barely used a few weeks later; employees who keep maintaining their old spreadsheets "just to be safe"; usage that exists on paper but gets worked around in daily practice. This quiet form of failure is considerably more common than open revolt – and considerably harder to spot before the budget is already spent.
Practice section: change management as its own project line item
The practical consequence of this number is easy to state but rarely implemented: change management belongs in every AI project as its own, budgeted component – not as an optional add-on, but on equal footing with the technical implementation. The following modules in this course show why middle management specifically plays an underestimated role, which employee fears are actually legitimate, and which concrete methods build acceptance systematically.