AI Fluency and the New Surface of Work
Habit changes imagination.
Have you ever been reading a paper book and caught yourself wanting to hit Control-F to find something? So much so that your fingers made the keystroke? (Maybe it's just me.) Or put two fingers on a printed image and tried to zoom in? Or opened a paper map, expecting a dot to show exactly where you are?
In psychology, this kind of error is called an action slip. Through repetition, familiar actions become so automatic that we sometimes perform them in the wrong context. Would that Control-F could find my keys, or resurrect that thought I had last week that is now gone forever.
If you're using AI regularly, you may recognize the beginnings of a similar habit. You start thinking in prompts rather than opening software. You think about the actions and architecture needed to get AI to do something for you, and where you need to step in, direct the work, and apply your own judgment. You begin to imagine the checkpoints you would build into a process, where your expertise needs to take the lead, and what you can safely leave to AI.
Habit can truly open the imagination. When we become proficient with tools, routines, musical instruments, recipes, we begin to see possibilities. The better we get, the more those skills influence what we imagine we can make. A programmer sees possibilities in code. A designer can imagine things that someone unfamiliar with design software may never consider. Someone who knows PowerPoint well can see a presentation taking shape before opening the application.
The tools we know become part of how we think. And as we become fluent in them, they change what we believe is possible.
Fluency changes how we understand work.
At this moment in AI, I believe we have two ways of approaching work emerging at once. People who have not yet developed AI literacy or fluency tend to think about work through its familiar structures. People who have achieved some level of fluency—who may not be experts, and who certainly don't need to be technical—are beginning to think about work very differently.
They may think of AI as the peer they never had. They think about data, direction, context, and the "skills" they want to develop and give to AI. They approach a problem by considering what they could build, automate, or ask AI to help them figure out. The output might be something they wouldn't previously have considered making.
These are not two fixed groups of people. Fluency increases the likelihood of approaching work differently, but it doesn't guarantee better thinking. Still, I see a growing difference in how people imagine the work in front of them.
"We shape our buildings and afterwards our buildings shape us."
Winston Churchill said this to the House of Commons in October 1943, arguing that the chamber, destroyed by bombing in 1941, should be rebuilt as it was. He wanted to keep its oblong shape rather than a semicircle, and his reason was behavioural. In a semicircle, he said, members drift around the centre as conditions change. In an oblong chamber, crossing the floor is an act that requires serious consideration; he spoke from experience, having crossed it twice. He wanted the two sides literally on two sides of the room, and not in a circle.
Traditional software is a building most of us have lived in for decades. It has shaped how we organize information, communicate ideas, and even how we think. And for a great many organizations, one piece of software has had an outsized influence: PowerPoint.
Open PowerPoint and you're invited to think in slides, shapes, and arrows. I get to see different organizations' "slide cultures" and what they reveal about how those organizations think. In universities, slide design is often secondary to the flow of ideas. Startups tend to think in pitch decks, with every slide making the case, layer on layer. Engineering cultures I've come to know seem to prefer PowerPoint over Word, and the bullets and levels of indentation approach the absurd. Step into a creative ad agency and the minimalism of a slide can be so extreme that if you didn't have the creative director speaking over the deck, you'd never know what it was about.
I'm never quite sure whether all that structure leads to better logic or just better pixel alignment. Which drives which—the shape of the tool or the shape of the culture?
AI is becoming a new culture-shaper, and its signature move is changing where the work begins.
When you start with a well-constructed prompt, well-built context, or simply a conversation with AI using your voice, you start closer to the problem. You can explore what you're trying to accomplish before deciding what form the work should take. AI can help you figure out whether the result should be a presentation, an argument, a working model, a website, or something else entirely.
We're seeing this in the tools themselves. AI systems are beginning to create and work directly with documents, presentations, code, websites and other outputs that once required separate applications and specialized knowledge. Video generation is perhaps the most spectacular example. But someone who knows video should still be a better director of AI-generated video than someone who doesn't. Fluency with AI doesn't eliminate the value of knowing the craft.
The more interesting development for most of us is happening in everyday knowledge work. The familiar stack of office software is beginning to change, and with it, the assumptions we've carried about how work gets done.
With AI, the shape of work no longer gets decided before the thinking starts, and that's what's fascinating and rewiring. It's a new work surface we've never seen before—outside of our own internal ways of thinking.
Organizations need to close that gap.
As AI becomes the new surface of work for more people, literacy and fluency become critical to developing judgment and differentiation.
Any technology has its own conventions, and creating that muscle memory gives us more room for exploration. We begin to understand what the tools can do, where they fall short, and how to work with them. We become more likely to ask "why this way?" and consider whether something could be approached entirely differently.
But there is a growing gap between people who are beginning to think this way and those who still understand work through the software and processes they've always used.
Imagine two colleagues working on the same problem. One thinks about the report that needs to be written, the research that needs to be gathered, and the slides that will communicate the findings. The other begins by asking what information is needed, what AI could investigate or build, and whether a report or presentation is even the best way to deliver the result.
Neither is necessarily wrong. But they may be imagining different work, even when they agree on the problem.
The best AI users are beginning to think beyond automating what they already do. They're asking what becomes possible when the old constraints are removed. What could we build? What could we offer? What problems could we solve that weren't previously worth attempting?
This is where organizational fluency becomes important. A common understanding of AI helps people compare ideas, question assumptions, and work together on possibilities that might otherwise remain confined to individual experiments. You begin to flatten the distance between the heavy user and the new user when both can speak the same language and understand what is possible.
Fluency changes what individuals can imagine. When that fluency becomes shared, teams can explore those possibilities together. Some of those explorations will change how the organization works and what it offers.
Automating workflows gets you speed; fluency is more likely to get you differentiation. Neither is guaranteed, of course. But the more possibilities we can imagine, the more opportunities we have to create something worthwhile.
This is not an overnight shift, although if you've had your three sleepless nights about AI, you might disagree. We're just starting to explore what's possible in most organizations, and beginning to question whether the structures of the past should be the structures of the future.
Once you see it, it's hard to unsee.
There is a bargain in this, as there usually is. Habits that open the imagination can eventually close it off. Eventually, the frame we're fluent in can become the only frame we think in, as the slide deck did for a generation of us. AI may help us question those habits, but it could just as easily reinforce them. We'll develop favourite prompts, methods and assumptions, and some will become second nature even as the technology changes underneath us.
I'll take that bargain for now and push for fluency that's not yet the norm. I'll worry in the future about the habits we'll eventually need to unlearn.
First, I'd like to discover what we can do with AI, and help our clients discover what they can do that they've never done before.
I'm the Head of Innovation and Strategy for Canada at NATIONAL Public Relations. I work at the intersection of change, culture and strategy, helping leaderhip teams and organizations think through what's next. I write about these and other things on LinkedIn and at KevinMcCann.ca.
This article was inspired somewhat by recently joining an Atlantic Economic Council Business Roundtable on AI adoption in Atlantic Canada. We discussed these imperatives: the need for leaders to get involved, more hands-on AI skills across the workforce, practical guidance to help businesses adopt AI safely, and a stronger focus on rethinking work and turning experimentation into business value. Which I think is pretty much reflected in this piece. Thanks David Chaundy and Lana Asaff for that invitation, and Josh Lee, Matt Symes , Andrea King , James Mosher , Tony Fang and others for that conversation.
Influences, references, and further reading
- Mollick, Ethan. One Useful Thing and Co-Existence. https://co-existence.ai/ (for consideration), thank you Ethan Mollick for your great work and excited for your new book.
- Cheek, Paul. (2026). No One Works Here: How AI-Driven Enterprises Are Raising the Bar for Business, Leadership, and Competition. Thank you Paul Cheek for our chat and your work!
- Most People (and Companies) Aren't Tool Builders. Podcast episode, Apple Podcasts. (for consideration), thank you Benedict Evans for your great work.
- AI and Communications: Practice, Leadership, Careers, Adaptation - an earlier essay I wrote.
- Communication Is How We Create Meaning - an earlier essay I wrote.
- Norman, Donald A. (1981). Categorization of Action Slips. Psychological Review, 88(1), 1–15.
- Churchill, Winston. (1943). House of Commons Rebuilding — Parliamentary Debates, October 28, 1943. UK Parliament, Hansard.