In IT, change control is a basic discipline. Before a major system change goes live, someone reviews it, tests it, documents who is affected, and keeps a plan to roll it back if something breaks.

AI is being rolled out across the world with very little of that. Vendors update their models every few weeks, often changing how a tool behaves without the customer approving or even noticing. Employees are handed tools that can shape hiring, lending, medical, and customer decisions, often without training on when to trust the output. Meanwhile, regulation varies widely by country and much of it is still being written, so there is little outside pressure for organizations to slow down and manage the change properly.

Supporters of a lighter touch approach argue that strict rules would slow innovation and let other countries pull ahead, and that concern deserves a hearing. But even without new laws, companies do not need permission to practice good change control. The organizations that test, train, document, and keep a way back will avoid a lot of the damage that comes from moving fast without a plan.

Some recent articles echo my sentiments (see below). Companies buy the tools, run a successful pilot, and hand out licenses. A few months later, only a small share of employees actually use them. I agree with the authors that the technology is rarely the problem. The problem is that most organizations spend months choosing a platform and only a few weeks preparing their people for it. ChatGPT.ca estimates that an AI rollout is about 20% technical and 80% organizational, yet most project plans are built the other way around.

WillDom’s Linkedin article makes a good point about why AI is harder than other software. A new CRM asks people to do the same job in a different screen. AI asks people to change how they make decisions and what they trust. It also raises a question nobody asked about the CRM: will this take my job? Employees rarely say that out loud, but it shapes their behavior.

They comply on the surface and quietly go back to the old way of working. Robert Half found that only about a third of workers feel very confident using AI tools. The rest are not against AI. They simply have not been shown how it fits into their day.

The most useful story came from Rework. A director at a logistics company ran three AI rollouts in two years. The first two followed the usual script of announcing the tool, scheduling training, and sending logins. Both stalled within 60 days. For the third, she changed only the sequence and waited to train people until they had a real problem the tool could solve. Adoption went from 14% to 71% in six weeks. In customer success, we know that adoption has to be earned one customer at a time. The same is true inside a company, where employees are the customers.

With change control as a high priority, where should an organization start?

• Involve the people who will use the tool before launch, and let them help define what success looks like.
• Address the job security question early and explain specifically what will change for each role.
• Teach judgment in training, meaning when to trust an AI output and when to question it, not just which buttons to click.
• Choose champions based on influence rather than enthusiasm. As Fronterio notes, the best champion is the person colleagues already go to with hard problems.
• Have managers and executives use the tools visibly themselves, because teams follow what leaders do more than what they announce.
• Act on employee feedback where people can see it.
• Measure actual usage for at least 90 days, not just logins.

An AI investment is only worth as much as the number of people who actually use it. Leaders planning an AI rollout this year should read these articles and give the people side of the project the same attention they give the technology.

Sources/articles:

https://chatgpt.ca/blog/ai-rollout-change-management

https://www.linkedin.com/pulse/change-management-missing-layer-most-ai-rollouts-willdom-js2qe

https://fronterio.com/en/blog/ai-change-management-enterprise-rollout-stalls

https://resources.rework.com/guides/ai-team-readiness/change-management-ai-rollout

https://www.roberthalf.com/us/en/insights/management-tips/ai-change-management-for-leaders