The Corporate Understudy

That job is being absorbed.

Not entirely, not yet. Clayton, who also carries RSC credentials and screen credits in Peep Show and Law & Order: UK, was interviewed recently on NPR and made the distinction carefully: AI can handle the purely instructional. If a company needs someone to explain a process, deliver compliance steps, demonstrate a procedure — the machine does it efficiently and for considerably less than a working actor. That lane belongs to the algorithm now.

What AI hasn't cracked, he says, is interactive training. The scenarios where the trainee has to respond, adapt, push back, misread, and try again. Where something resembling genuine presence is required on both sides. That's where actors still have purchase — the messy, unpredictable, can't-fully-be-scripted quality of human interaction in high-stakes moments.

This is presented as good news. And in terms of employment it is, marginally.

But there's a more uncomfortable reading, one the interview doesn't dwell on.

The corporate training video is one of the stranger artifacts of modern organizational life. Companies spend real money staging dramatizations of what they claim to value — the right way to deliver feedback, the correct response to harassment, what a psychologically safe conversation actually looks like — and then show these staged versions to their employees as aspirational templates. The actor is hired to perform the values the company nominally holds. Sometimes those values are practiced. Often they're not. The training video becomes a liturgy: official doctrine recited in a conference room, separate from the lived reality of the building.

The actor in the scenario isn't just performing. They're the organization's designated honest moment — the one place where the difficult thing is named out loud and rehearsed. The awkward manager, the boundary-crossing colleague, the HR rep who actually helps. These characters exist in the video with more clarity than the humans in the building often manage.

Now AI is auditioning for that role.

Here's the part worth sitting with: if AI can perform it well enough for the purely informational version, the question isn't about what AI can do. It's about what the performance was doing in the first place. Because if the point was ever to actually change behavior — to give people a real map for navigating difficulty — then whether a human or a machine delivers it was always secondary to whether anyone was watching to learn, or just watching to check the box.

What Clayton's career reveals, inadvertently, is that we've been outsourcing the performance of self-awareness for decades. We hire actors to demonstrate what we ought to do, file the completion certificate, and the machinery of the actual workplace continues. The training video is the mask the organization wears to itself.

That's not Clayton's failure. He's bringing craft to a real job, and his instinct that presence matters in interactive training is almost certainly correct. The discomfort doesn't belong to him — it belongs to the gap between what training is supposed to accomplish and what it reliably does.

AI stepping into the informational lane doesn't close that gap. It makes it more visible: that we've been treating the performance as the substance. Now that we can generate the performance cheaply, we're discovering we may have been doing that all along.

i · sources

• Paul Clayton discusses his role in corporate acting as AI characters gain popularity — NPR / WKNO FM, 2026-05-23