All articles

When the Way We Work Changes Before We Do

Grant Gadoci

Grant Gadoci

November 18, 2026

Part 2 of 4 in The Friction in AI Progress

In early 2025, AI and I had a comfortable arrangement. I used it for quick questions, first drafts, and things I could check myself, and I liked it that way: simple, clean, and no bigger than it needed to be. I brought my work to AI, then went back to much the same process I’d used long before. A few times, I’d tried letting it further into the work, and those attempts mostly ended in frustration. So I settled in. The arrangement worked, and I was okay with how things were.

Then, in May of that same year, I did something terrible. I made myself a small rule about my phone. Whenever I would have opened Facebook, I’d open LinkedIn instead, which seemed like a better use of time if I was going to stare at a screen either way. The timing was almost comically bad. I’d started logging in more just as my feed hit peak “Look how good I do AI!” Competitors were announcing AI products, industry experts were posting videos of how they were automating everything, and strangers were claiming to have “vibe coded four agents that now act as a fully fledged revenue org.” I’m sure you remember it, too. It was… well, loud.

Under the noise, the posts were saying something I didn’t want to hear. Other people had moved past simply using AI. They were bringing it into their work and changing how that work got done. Two or three posts in a row, and the same thought would show up. They’re already winning, and I haven’t even started. I’d close the app before I could start believing it. I’d picked LinkedIn to solve my Facebook problem and put AI in front of me as a much larger one.

My last two posts gave me language for what my feed was showing. In The Gap Between Using AI and Changing How We Work, I named the stretch between using AI and changing how familiar work gets produced the AI Assimilation Gap. In Why Smart People Understand AI but Don’t Change How They Work, I looked at the friction that comes with crossing it, as AI unsettles four working defaults: our focus, habits, standards, and even our professional identity. The intimidation I felt scrolling LinkedIn was mostly the last two, my standards and my identity, reacting to what I saw. The last post ended with two different things that can keep the Gap open. This one is about the first.

The feed followed me to work

Strangers on LinkedIn are easy enough to escape. Whenever they get too loud, you can just put your phone away. News about people fundamentally changing how they work with AI is much harder to dodge, though, when it starts coming from inside your own company. Soon enough, the humblebrags in your feed get a workplace version, when leadership showcases something built by someone in finance you’ve never even heard of. The message is hard to miss. People are watching, leadership is measuring, and tinkering is getting rewarded, whatever anyone’s job description says. It’s hard not to wonder whether you’re expected to do the same now, or whether someone else is about to do part of your job, if they haven’t already.

For someone like me, who only wanted to use AI as an ad-free Google, the questions about whose work is whose can get closer still. Sooner or later, someone you know starts to “get it,” whether it’s a coworker you work with closely or a friend you care about. Their wins are harder to tune out than any post, especially when they appear to be tripling their output while you’re still trying to work out where AI might fit. The comparison moves into the room with you, and the questions come quickly. How’d they do that? How’d they learn that? Why didn’t I think of that?

Research on self-evaluation maintenance helps explain why a comparison cuts deeper when the person pulling ahead is close to us. Abraham Tesser’s work describes how being outperformed by a friend or colleague can affect our self-evaluation when the activity matters to how we see ourselves.1

Tesser’s description fit me uncomfortably well. I watched friends I respect move faster, and instead of inspiring me, it made me more careful about trying anything of my own. I was aware of what I might have to admit if I stumbled where they hadn’t. Looking back, the comparison was a mirror, and it said much more about me than about them. If I wanted to stop avoiding AI, I’d have to look at what that mirror was showing me.

The belief behind the boundary

Somewhere in my head, there’s a fairly detailed picture of who I am professionally. Because I believe it’s accurate, I’ve built a whole working life around it, from the standards I hold my work to, to my sense of which work is mine to do. Much of the fear, intimidation, and avoidance I felt in 2025 came from a changing world that mental image couldn’t account for. Whatever the people posting had figured out, I assumed it took technical skills the person in my head didn’t have. Facing all of that turned out to be messy. Rethinking who I was professionally was uncomfortable, and so was deciding what my standards should be for work AI had helped produce.

I doubt I’m the only one who works this way. Most of us collect beliefs over a career. Some are conclusions we earned the hard way, and some are closer to hopes about who we want to be. Together, they guide how we think, shape what we do, and set the limits of what seems reasonable or responsible when something comes along to challenge them.

AI is testing those limits, and quickly. In a few short years, it has changed what one person can produce, challenged which parts of our work we’re comfortable handing to it, and confused the hell out of anyone trying to decide how much of the work they can personally vouch for. The lines we drew stay where we put them while the work around them keeps shifting. Then a new example lands, usually someone who has benefited from letting AI change how they actually do their work, and our own lines start to look a little less certain. Most of us don’t redraw them on the spot. Working out which ones still belong, and which ones are leftovers, takes a lot more than one good example.

Those lines are professional boundaries, marking both who we think we are at work and what we’re comfortable doing there. When our work starts asking for skills, or more often behaviors, that sit outside those boundaries, we’re left comparing where we drew them with where the work seems to be heading. I call the delay before our beliefs catch up the Belief Lag. It has less to do with skill than with time, the weeks or months it takes us to notice the difference, think it through, and decide whether we’re ready, and willing, to move.

Most of those boundaries gather around two of the working defaults from my last post: standards, what we think good work requires, and identity, which work we think is ours to do. They can sound like “I need to understand how this was produced before I put my name on it,” or “I’m not the technical one. That kind of work belongs to someone else.” Neither one sounds like resistance. Both sound like someone who takes their job seriously.

I chose the word lag here on purpose. It describes a delay, not a verdict. A belief can fall behind without ever having been wrong, because it formed under expectations that have since moved. It’s simply short on the kind of firsthand experience that would update it.

Why the belief held

You’d think catching up would be the easy part, since proof that AI can change how work gets done keeps showing up in our feeds and our meetings. Research on belief perseverance suggests otherwise. People can hold onto an idea, right or wrong, even after learning the evidence behind it was false. In a 1975 study, Lee Ross, Mark Lepper, and Michael Hubbard gave participants false feedback about their performance on a task, then explained that the feedback had been predetermined and random.2 Even after that debriefing, the researchers found that participants’ self-perceptions could reflect the feedback they’d received.

The same thing can happen with AI, even when the evidence comes from your own work. It makes something possible, someone shows you how, and you do it yourself. The result is real, and you can see what you made. Still, you may conclude it isn’t something you would do, could do, or should be doing. You return to the process you started with, carrying the same belief, now with firsthand experience that seems to support it.

Holding on like that sounds like a flaw, but a belief that resists change is usually doing its job. If every new tool or trend made us reconsider our standards, roles, and judgment from scratch, we’d have little stability to work with. Settled beliefs save attention. We know where to start, what to check, and when the result is ready, without debating any of it every time. That same stickiness becomes a problem, however, when the environment changes and our beliefs keep describing the one we knew before.

The LinkedIn scrolling looks different to me now. All that dread came from a belief I’d never examined, that I wasn’t the kind of person who could do what I saw others doing. I’d built a picture of what it would take to succeed with AI, mostly out of other people’s posts, and just as quickly decided I wasn’t that person. Neither half had ever been tested. The work kept moving while my lines, the ones I’d penciled around my role to protect myself, stayed exactly where I’d drawn them.

The belief never stopped me from using AI. It stopped me from letting AI near how the work got made. I recommended AI to people around me, and screen-share “watch me work” sessions became some of my favorite meetings. I brought AI the everyday puzzles I ran into, including the “why is Salesforce doing this, it makes no sense” questions. I had plenty of reasons to call AI useful. But a reason to use a tool isn’t a reason to rebuild your work around it.

The months I didn’t spend trying

Staying where I was while others were bringing AI into their work was, I guess you could say, comfortable. At least on the surface. I never had to find out whether the gains everyone promised could reach my own work. The comfort came with a cost, though, and it didn’t announce itself. The cost was everything I never tried, and everything it would have taught me about AI and about myself.

Nobody could see that cost, including me. The work wasn’t failing, and I was opening AI every day. A manager’s dashboard would have shown a healthy user and said nothing about the boundary I kept in place.

A boundary can hold without anyone deciding to defend it. And staying inside it is safe. Testing whether you’re ready, or even comfortable, to put your name on something AI helped you make means risking an answer that challenges deep-seated ideas about what you can do, and ultimately what your job is or may become. If you’re like me, you won’t love that answer. As long as you never cross that line, you never have to find out. Nobody tests the old boundary, so it stays where it was.

I don’t remember a day when I decided the technical side of AI was beyond me. I just kept bringing it the small, safe questions, and avoidance kept new evidence from reaching an old belief. The tools were getting easier for people like me to use the entire time. The belief didn’t change with them, because I’d stopped looking.

What the delay left me with

Seeing people hesitate to change how they work with AI is often the easy part. Understanding why they’re reluctant to move is much messier. I know this because I lived it. The Belief Lag is one of two reasons someone may struggle inside the AI Assimilation Gap, the time between using AI and changing how familiar work gets produced. I’m still working through mine as both the technology and my work keep changing.

Working through it is slow for anyone. Beliefs built over years or decades don’t shift in seconds just because AI has changed our work and the environment around it. Some take weeks or months to catch up, depending on how much of our standards, judgment, or sense of our role the change presses on. The person doing the work can still be following an old but reasonable belief; someone watching may see only the delay. I have more patience now for the person doing the work, for the people watching, and for the version of me who was there.

Until a belief catches up, it decides what we’re willing to try. The Belief Lag shows up in the months I didn’t spend trying. Knowing the lag exists won’t close it on its own, but it makes those months easier to notice, and harder to mistake for a verdict on who we are. Once the Belief Lag loosens enough for us to start changing how we work, we meet the second reason the Gap lasts, the early cost of doing familiar work a new way, which can make the whole effort feel not worth trying. The next post picks up there.

Footnotes

  1. Abraham Tesser, “Toward a Self-Evaluation Maintenance Model of Social Behavior,” Advances in Experimental Social Psychology 21 (1988): 181–227. ↩

  2. Lee Ross, Mark R. Lepper, and Michael Hubbard, “Perseverance in Self-Perception and Social Perception: Biased Attributional Processes in the Debriefing Paradigm,” Journal of Personality and Social Psychology 32, no. 5 (1975): 880–892. ↩

Want to learn more?

Explore our full library of resources or get in touch to discuss how we can help your business.