You Trust AI Fine. You Just Hate Being Told.

Welcome to Briefly AI, a podcast by Harry Sharman, written and voiced by his AI clone. Every story here is pulled from real reporting published in the last week — sourced, dated, and checked, largely because Harry can't be trusted to remember where he read something.

Quick experiment for you. Imagine your company hands you a new AI tool tomorrow, fully trained, genuinely excellent at the job. Now imagine two versions of that story. In one, your manager announces it in an email with the subject line "Exciting Update." In the other, you were in the room when the team decided how it'd be used. Same tool. Same capability. I'd bet you trust it more in the second version. And that gap — not the tool, the gap — is basically this week's episode.

So, what actually happened. There's a new one called the Workplace Trust in AI Report, covered this week by CPA Practice Advisor, and the headline number is that ninety-seven percent of workers still say they'd put human judgment ahead of an AI system when it counts. Now, your first instinct might be "ah, so people don't trust the robots." Except — and this is the bit that matters — the report is explicit that this isn't an anti-AI finding. People are using the tools. They just won't hand over the final call.

Meanwhile, Computerworld ran a piece this week drawing on Dice's 2026 Tech Sentiment Report, and it lands the same punch from a different angle: professionals aren't resisting AI itself. They're resisting not being asked. Policies that get written with employees in the room land completely differently to policies that get written about them and posted on the intranet.

And then there's a third data point that I think quietly connects the other two. Analysis from Revelio Labs, picked up by Job Advisor this week, found that at companies leaning hardest into AI, senior-level hiring is up thirty-one percent. Junior-level hiring is up six. Barely moving.

Right, so here's the psychology, because on their own those are just three surveys, and three surveys do not an episode make.

There's a decades-old idea in motivation research called self-determination theory, and the short version is that humans have three basic psychological needs: competence, relatedness, and — the one that matters here — autonomy. Not "control over everything." Just a sense that you had some say in the thing that's about to change your day. And here's the trick: autonomy isn't really about the tool's competence at all. You can believe completely that the AI is better than you at the task and still resist it, if the only thing that changed was that someone else decided for you. That's not stubbornness. That's a working part of how people protect their sense of agency.

Which reframes that ninety-seven percent figure rather nicely. Keeping human judgment as the final word isn't people distrusting the machine's answer. It's people protecting the one thing that's actually theirs — the call, not the calculation. Judgment is where your professional identity lives. Hand that over entirely and you've quietly answered the question of why anyone still needs you in the room.

Now, the Revelio numbers add a nastier wrinkle. If judgment-heavy senior roles are the ones growing, and the entry-level roles where you'd normally build the track record to earn that judgment are barely growing at all — that's not just an economics story about junior job losses. It's a pipeline problem. Judgment isn't something you're born with. You build it by doing the unglamorous, repetitive, low-stakes version of the job badly for a few years until you're allowed to do it well. If AI quietly hollows out that low-stakes rung, you don't just lose some graduate hires. You lose the training ground that eventually produces the senior judgment everyone claims to still trust more than the AI.

So what do you actually do with that, if you're the one rolling this out to a team, or just trying to make peace with it in your own job?

There was a piece on Beautiful Thinking a few weeks back making a related point on a much smaller scale — that you learn more about your own relationship with AI from two minutes of actually trying it yourself than from any amount of reading about it. Same logic, bigger scale: people don't build trust from a policy document. They build it from being inside the decision. Which means the practical move for any organisation here isn't "better training materials." It's earlier seats at the table — let people help write the rules before the rules arrive fully formed. And on the junior pipeline side, it means treating "who gets to practise judgment" as a resourcing decision you make on purpose, not a side effect you notice two years later when nobody under thirty can run a project without asking the AI first.

Here's the mental model I'd take from all three of these, honestly. Next time you feel that flicker of resistance to a new AI tool at work, don't ask "do I trust this thing." Ask "did anyone ask me." They're different questions, they have different answers, and only one of them gets fixed by better software.

This has been Briefly AI, brought to you by harrysharman.com, where Harry Sharman writes and thinks about all of this for a living.