Why 90% AI Adoption Isn't the Win HR Thinks It Is

Editor's note: Posts published on From the Workplace are written by outside contributors and do not necessarily reflect the views or opinions of SHRM.

An HCLTech study of 500 enterprise decision-makers found that 90% of organizations say generative artificial intelligence is transforming their workflows. Only 18% say AI is actually delivering revenue impact.

That gap should worry every HR leader more than it currently does, because it's not really an AI problem. It's the same problem learning and development (L&D) has had for decades, just showing up faster and more visibly now that AI has sped everything else up.

Specialty Credential: AI+HI

I spent more than 18 years as a Director at Deloitte and PwC, and one pattern showed up in nearly every organization I worked with: People would sit through the training, nod along in the room, rate the session highly on the feedback form, and then go back to doing exactly what they were doing before. This just proves that knowing something and actually changing your behavior around it are two entirely different skills, and most corporate learning is only built to deliver the first one.

Comparatively little of what people learn in training actually results in lasting behavior change. That situation predates AI by decades. What AI has done is make the gap more visible and more expensive, because now organizations are pouring real budget into tools that promise transformation and mostly deliver adoption. Employees log in. They use the chatbot. They complete the module. None of that is the same as the organization actually working differently six months later.

The HR leaders who will close this gap aren't the ones buying the most AI tools. They're the ones asking a harder question before they buy anything: does this actually change what my people do on a Tuesday afternoon, or does it just change what they consume?

A few things worth checking before the next L&D investment gets signed off:

• Does the tool personalize the action, not just the content? A generic recommendation delivered efficiently is still generic. Real behavior change requires knowing where a specific employee actually is, what they're actually stuck on, and what their next concrete step looks like, not a one-size-fits-all module.
• What gets measured after week one? Completion rates and satisfaction scores tell you almost nothing about behavior change. The better question is whether anyone checked in on whether the new skill or habit actually showed up in someone's day-to-day work a month later.
• Is the learning a one-time event or an ongoing loop? Behavior change rarely survives a single session. It needs reinforcement, follow-up, and some mechanism for catching people when they slip back into old patterns.

The gap between the 90% of organizations using AI to transform their workflows and the 18% seeing revenue results isn't a sign that AI doesn't work. It's a sign that most organizations are still buying for the wrong outcome. Adoption was always the easy part. It always will be. The harder, more valuable question is whether anything is actually different because of it.

Submit Your Article

Shruta Satam is co-founder and CEO of Pustakh, an AI-powered applied learning platform, and previously spent more than 18 years as a director at Deloitte and PwC advising large organizations across the US and Australia.