The most consequential decision in your AI rollout will not be made by your CIO. It will be made by whoever writes the words in the announcement.
I have thought about that a great deal since April, when I took a call from the Chief People Officer of a global engineering firm. She was in Europe, I was in Chicago, and she had a draft people strategy on her desk that had to go to the executive committee the next morning.
She did not want to talk about the strategy. She wanted to talk about a word.
“Do we use the word innovation, or do we use the word digital, or what is it we’re asking?” she said. “Is it too soon to say that we need digitally enabled leaders? Or is it the right thing to say?”
Then, about forty minutes in, she asked the question underneath all of it. “I feel like I don’t want to use AI. Is that wrong of me?”
I told her I wanted to think about it before I answered, which is not something I say often.
She was not hedging and she was not being squeamish. She was doing the most consequential piece of work available to her that week, and she was one of the very few executives I have met who understood that. The words in a people strategy are not the wrapping around the strategy. For the thousands of people who will never read the board pack, the words are the strategy.
The word carries a cost
The evidence is cleanest outside the organization, so start there.
Researchers at Washington State University ran experimental surveys with more than a thousand US adults across eight product and service categories. Same product, same description, one difference: whether “artificial intelligence” appeared in the text. The version mentioning AI performed worse every time. Lead author Mesut Cicek found that mentioning AI lowered emotional trust, and lower emotional trust lowered purchase intent. The effect was strongest for high-stakes purchases, the ones where being wrong costs you money or safety. His advice to marketers was blunt: describe the features and the benefits, and leave the buzzword out.
Inside the organization it matters more.
Slack’s Workforce Index surveyed 17,372 desk workers across fifteen markets, Australia included. Forty-eight percent said they would be uncomfortable telling their manager they had used AI for a common workplace task. The three reasons they gave, in order: it feels like cheating, they fear being seen as less competent, they fear being seen as lazy. The same research found that workers who feel comfortable discussing AI with their manager are 67 percent more likely to actually use it.
So half your workforce is already using these tools and hiding it from you, and the strongest predictor of whether they use it in the open is whether the language around them makes it safe to say so.
Their fear is rational. A study published in the Proceedings of the National Academy of Sciences in May 2025, four preregistered experiments with 4,439 participants, found that people who use AI at work are judged by colleagues as lazier, less competent and less diligent. Not their output. Them. Hiring managers penalized candidates who disclosed AI use.
But the researchers, Jessica Reif, Richard Larrick and Jack Soll at Duke, found two things that made the penalty shrink. It shrank when the evaluator used AI themselves. And it shrank when the tool was clearly described as useful for that specific task.
The stigma is not attached to the technology. It is attached to the ambiguity around it. When the organization has said plainly what the tool is for and why it belongs in this piece of work, the judgment largely evaporates. The language your leaders use is not communications support for the rollout. It is a control on whether the rollout works.
Six rules I would give any executive team
Name the task, not the technology. “We are cutting the time it takes to produce a compliance report from four hours to forty minutes” lands. “We are becoming an AI-enabled organization” does not, because nobody can picture themselves inside it. Specificity is what removes the stigma.
Say what stays human in the same breath. When an accountant of thirty years hears that AI will handle the monthly reconciliations, the question is not only whether they still have a job. It is who they are now. Purpose and identity are what this technology confronts, and no other change program in my career has done that. If your announcement does not answer “what is my work now”, you have announced a threat.
Retire the words that have gone soft. Digital transformation. Digital disruption. Innovation. One of her executive committee members had told her the firm does not really innovate, not in the patented, trademarked sense. They solve complex problems for clients. Using a word your own people quietly know is untrue costs you more credibility than saying nothing.
Be precise about “digitally enabled leadership”. I think you can and should say it. But it has two halves, and if you only say the first you have said something frightening. It means leaders who use the technology to do their work better. It also means leaders with the human skills to lead people through a period when the technology is changing what their work is. Both, out loud, every time.
Never let AI take the blame for a management decision. Cost cutting dressed up as AI strategy has a name now, AI washing, and your workforce can smell it from a distance. It buys you a quarter of investor sentiment and costs you years of trust.
Choose augmentation language only if you mean it. Writing in Harvard Business Review in April this year, Jan-Emmanuel De Neve of Oxford, Jeffrey Hancock of Stanford and Kate Niederhoffer of BetterUp argued that automation strategies show the faster early gains but augmentation wins over a longer horizon, because long-term performance depends on how people feel about their work, whether they genuinely engage with the tools, and whether the good ones stay. Using augmentation language over an automation strategy is not clever positioning. Your people work out the difference within a year, and then you have lost both the strategy and their trust.
The only test that matters
Late in our call she described the test she applies to every line of the strategy. Could she stand in a room of a hundred engineers who do this work all day, say this is what we are going to do for you, and have them nod.
Not whether the language survives the board. Whether it survives the room.
Your people are not waiting for you to settle the wording. She said as much to her own executive committee: they are already adapting, already changing how they work, already solving it without permission, while leadership figures out who is in charge. The language question is not about controlling them. It is about whether what they are already doing happens where you can see it, shape it and learn from it, or somewhere you cannot.
So pick the words with the same care you gave the platform decision. They are doing more work than you think.
