Most of the leaders I've spoken to this year have bought AI. They have the licences, the consumption bill and the internal comms to prove it. What quite a lot of them don't have is a business that works differently to the way it worked eighteen months ago.
The tooling is genuinely good now. The direction of travel is clear. And there's still next to no ROI showing up inside most organisations. That gap, between what these tools can do and what companies are getting out of them, has almost nothing to do with technology.
Quite a few organisations bought AI without ever writing down the problem they were buying it for. The board asked what the AI strategy was, and the answer had to exist, so it got bought. We've done this before. When big data arrived, everyone needed a data lake, and a lot of companies ended up with beautiful infrastructure and no answer to which decision it was going to improve. The gut instinct still made the call.
That costs you the one thing that would move adoption. If you can't name the problem, you can't tell your people why their work has to change. And a workforce told to adopt something without a reason will supply the reason themselves. The reason they land on is the one that ends with their job going.
Resistance to AI usually gets diagnosed as irrational, as though people were being sentimental about spreadsheets. The fear has a logic to it.
Nobody at work is deluded anymore about what happens when an efficiency saving gets found. People have read the announcements: large cuts at profitable companies, in quarters where the results looked strong. They watched both sides of the pandemic. During lockdown the message was to look after your health and mind your family. Quite quickly after it, the message was that margins were tight and everyone was needed back in the office. And they can hear the technology companies telling investors that those valuations rest on capturing the work people currently get paid to do.
So when leadership says this is about making your job easier, the workforce does the maths. Most adults have handled quite a lot of change in their lives. What they handle badly is change with their own survival attached, and fear drives more workplace behaviour than any of us will admit to.
The dashboards on most of these programmes count licence uptake, active users, logins, prompts per week. They tell you the licence has been used. They tell you the login worked. They tell you nothing at all about why nothing has changed.
Talk to people and you find out which problem you're really dealing with, and there are only two worth separating. A team that's engaged and getting nowhere has a literacy gap. Show them what the tools do beyond summarising email and writing PowerPoint, and it moves. A team that says all the right words and volunteers nothing has decided the exercise isn't safe. No training budget will touch that one.
Both get reported upwards as adoption. They need completely different responses, and you can only tell them apart in conversation.
When a workforce won't come with you on something this big, I think the useful question is one for us. What have we done, over years, that means our people won't trust us with it?
Most organisations have spent that credit already. Every town hall where the numbers were reframed, every restructure sold as an opportunity, every invitation to speak honestly that ended with the problems being handed back to the people who raised them. All of it draws down the same account. Then AI arrives, needing more trust than any change programme in decades, and the account is empty.
What sits underneath is culture and systems, built up over years. Comms doesn't reach it. Training doesn't reach it. And it now stands directly between a large investment and any return on it.
The instinct is to walk into the room with a plan and a promise about redeployment and new kinds of roles. I've been guilty of selling that story myself. It's a hard one to sell now, because the direction of travel points at fewer roles, and at roles changing beyond the skills of the people currently in them. Everyone in the room knows it.
What works is saying what you actually know. I stood up in an all hands around the first lockdown, with furlough being announced and no idea where any of it was going, and said exactly that. Honesty was easy then, because nobody expected me to have the answer. The future is written in pencil right now, and admitting that costs very little. No employee is going to fall over in surprise at a leader who says they don't know how this ends.
The fear underneath the corporate language usually belongs to the leaders. We think we're supposed to have the answer. What people need is evidence that the person in charge is trying to protect them and the company at the same time, and will say so plainly when those two things pull against each other.
We've built our whole sense of professional worth on output. I'm valuable because I produced this. When the producing is increasingly handled elsewhere, what's left is how we behave: how we work a problem together, how we communicate, how kind we are with each other. I don't know how you'd build a metric for that, and I'm no longer sure you should. That may be the part of work we stop measuring and simply accept.
The trust that makes any of this work can't be bought with a licence and it won't show up in a dashboard. It gets built in how leaders talk to their teams on the days when there's no answer available. That work is vital and it can't be skipped. Do that, and the tooling has a chance. Skip it, and you're paying for software your people have every reason to leave alone.