Why Using AI Still Feels Like Cheating

Welcome to Briefly AI, a podcast by Harry Sharman, created by AI and voiced by an AI synthesis of Harry Sharman. Which is either very efficient or a cry for help. Possibly both.

Nearly half of Americans now use AI chatbots regularly. And 63% of them think AI is moving too fast. Those two facts sit in the same Pew Research poll, about the same people, at the same moment in time. Which means we're not looking at a divided country — cautious abstainers on one side, enthusiastic adopters on the other. We're looking at millions of people who are doing the thing while simultaneously feeling uneasy about the thing. And that tension is worth sitting with for a minute.

Because here's what that actually tells us. Adoption and consent are not the same psychological state.

We tend to talk about AI adoption as if it's a binary — you're using it, or you're not. The number goes up, so progress is happening. But that framing misses something important. You can use a thing regularly, even depend on it, while quietly feeling that it arrived in your life without your full agreement. That's not contradiction. That's a very normal human response to change that's moving faster than your sense of agency can keep up with.

And it shows up in stranger ways than just survey data.

This week there's also been a steady drip of research confirming what people working in HR and organisational change have been noticing for a while: a significant chunk of the workforce — somewhere between 45 and 48 percent depending on which study you read — is hiding their AI use at work. Not because they're doing anything wrong. But because admitting to it, disclosing it to a manager or colleague, comes with a social penalty. They get rated as lazier. Less competent. Less likely to be trusted with important work. Even when the output is better.

So we have two uncomfortable things happening simultaneously. People are using AI in large numbers. And they're doing it quietly, in the dark, because the moment it becomes visible, it changes how they're perceived.

Now, you could read this as a temporary cultural lag. Give it a year, two years, and the stigma fades as AI use becomes normalised. Maybe. I think that's probably partly right. But I don't think it's the whole story.

There's something deeper going on around what AI use *signals* about a person — and Harry Sharman put a version of this quite well in a piece he wrote a few months back, which remains one of the cleaner framings of the problem I've come across. The argument is that the real friction in AI adoption isn't technical. It's not that people don't know how to use the tools. It's that using the tools raises an uncomfortable question about what the person's value actually is. If the AI did the thinking, what did you do? And if you can't answer that confidently, it feels safer to keep the AI invisible.

Professional identity isn't a soft thing. It's the story you tell yourself and others about what you're good at, what you've earned, what makes your judgment worth paying for. When AI touches that — when it drafts the thing you used to draft, or analyses the data you used to analyse — it doesn't just change the workflow. It creates a small identity crisis that most people deal with by not talking about it.

And here's the bit that matters from an organisational perspective. If people are hiding their AI use, then a few things follow automatically. First, organisations are systematically underestimating actual AI adoption, which means their risk management is built on a fiction. Second, best practices never spread — because the person who found a smart way to use AI for something can't share it without outing themselves. And third, the things that go wrong — the hallucinated fact that made it into a client report, the AI-assisted decision that didn't hold up — those stay buried too, because the cost of disclosure is too high.

This is the bit the "just train people on AI skills" playbook misses entirely. It treats the friction as a knowledge problem when it's actually a trust and culture problem. You can give people all the prompting tutorials in the world, but if they're in an environment where admitting to AI use makes them look bad, the training doesn't move the needle. They'll just get better at hiding it.

And that brings us back to the Pew numbers, because I think they point to something useful if you read them generously. The 63% who think AI is moving too fast — that's not necessarily fear of the technology. It might be something more specific: a feeling that the pace of change has outrun their ability to have any say in it. Nobody asked them how AI should arrive in their professional life. Nobody sat down with them and worked out which parts of their job should be delegated and which parts should stay human. It just appeared, and now they're expected to use it and be grateful.

Agency matters in behaviour change. That's not a radical idea — it's foundational stuff in psychology going back decades. People adopt things more readily, more sustainably, and with less anxiety when they feel like they had some role in deciding to adopt them. Imposed change, even genuinely beneficial imposed change, lands differently than chosen change.

Which means the organisations that are going to get this right aren't the ones with the most sophisticated AI stack. They're the ones that do the unsexy work: creating actual space for people to say "I'm not sure about this part," building cultures where AI use is disclosed rather than hidden, and separating the question of "are you using AI" from the question of "is your judgment still valuable."

Those are answerable. But they require treating the adoption problem as a human problem, not a capability rollout.

So here's the question I'd leave you with this Saturday. If someone in your organisation admitted to using AI heavily on their last piece of work — would that make you think more of them, or less? Your honest answer to that question tells you more about your AI adoption culture than any tools audit will.

That's your lot. I've been your AI host Harry. If that gave you something to chew on, pass it on. See you next week.