Subversive Ventures
← Uncommon Sense
Field Notes

AI in the Boardroom: Talk is Easy, Adoption is Hard

24 Oct 2025David Ashenden
AI in the Boardroom: Talk is Easy, Adoption is Hard — AI Generated Illustration

AI Generated Illustration

It seems that most of the chats I have with board types start the same way: "We’re doing AI."

Usually followed by a proud smile, a consultant’s invoice, and a new tool nobody quite trusts. Then you talk to the team and find them quietly doing things the old way because it’s quicker, safer, and doesn’t crash halfway through a report.

It’s a bit like buying a self-driving car, but keeping your hands on the wheel because you don’t fancy testing its confidence on a roundabout.

Adoption is a trust exercise. People know their jobs, they know the little shortcuts that make things work. Suddenly you drop in an AI system that can’t handle exceptions, and the emotional cost of every wrong output outweighs the promised efficiency.

It’s also worth remembering that AI won’t solve the most nuanced problem in your business. It doesn’t do office politics, client nuance, or that sixth sense a good manager has when something feels off. But it can enhance almost any process that’s well defined. If it follows a pattern, AI will spot it, speed it up, and make fewer mistakes. The trick is picking the right work to trust it with, the kind that benefits from precision, not personality.

McKinsey calls it the "execution gap". I’d call it the Monday morning after the PowerPoint.

The firms quietly getting it right are the ones who start small and fix real irritations. Johnson & Johnson found that out of hundreds of AI pilots, only a handful delivered most of the value. Those wins weren’t glamorous, but instead practical. Automated checks, cleaner data, less admin noise. The stuff that frees up time rather than creates new meetings.

To build trust you need one reliable result after the other. Not slogans, grand launches, or transformation campaigns. Just tools that make the day run smoother.

And when board ask when the ROI will appear, it's the moment your team stops saying

"I’d rather just do it myself."

AI needs a job people actually want it to do more than grand visions.