I was half-watching Ron’s Gone Wrong with the kids for the umpteenth time when an article stopped me mid-scroll.
People in Los Angeles are strapping cameras to their heads and hands and filming themselves washing dishes. Making coffee. Folding laundry. Eighty dollars for two hours of footage. The footage trains humanoid robots.
Your first thought is probably the same as mine. Oh.
Because we’ve spent years reassuring ourselves. The skills AI cannot replace are the physical ones. The making. The hands. The workshop. Not the spreadsheets and the admin. That was the safe bit. That was ours.
And now we’re watching that story quietly collapse.
This is the conversation about what skills AI cannot replace that nobody in education is quite ready to have. The cameras are rolling in Pasadena kitchens. And I think we need to start asking what that actually means for the students in our rooms.
The story we’ve been telling ourselves
Because we have been telling ourselves a story, haven’t we. The story went something like this. AI will come for the admin jobs first. The writing, the coding, the spreadsheet work. But the trades, the making, the hands-on stuff — that’s safe. That’s the human bit. Our practical-minded students, the ones who light up in the workshop, they’ll be fine. The robots can’t do what they do.
And now we’re reading about people in Pasadena earning $1,200 a month teaching a machine how to chop vegetables and grill chicken. Wrist cameras capturing every muscle movement so a robot can learn the exact angle and pressure of the knife.
This isn’t just happening in LA. It’s happening in Nigeria, in India, in dozens of state-run centres in China. According to an LA Times article, entire families are signing up because the gig pays and the bills keep coming. Goldman Sachs reckons the humanoid market could be worth $38 billion by 2035. Investors put over six billion into humanoid robots last year alone. The people recording their kitchens are getting a tiny fraction of what’s being built on top of their data.
What the cameras can’t see
Here’s where my teacher brain kicks in.
We can’t keep preparing students for a world that’s already dissolving. Fact recall is finished as a goal. We knew that two years ago but a lot of curriculums haven’t caught up. Now it’s not just the knowledge economy. The motor skills economy is being captured too. Hand movements. Context switching. The small human adjustments we thought only people could make.
But I don’t want this to read as doom. Because the article also tells us something useful by what it leaves out.
What those cameras can’t capture is the decision layer. Why this solution and not that one. Who is this actually for. What trade-off am I making and is it worth it. When to stop iterating. When to throw the prototype away.
The motor execution is being recorded. The judgement isn’t.
That judgement layer is where the skills AI cannot replace actually live. And it’s the layer most assessment frameworks still treat as a footnote.
What this means for our classrooms
That’s our ground. That’s where D&T, art, every subject that lives in ambiguity, actually holds up. Not because the making is safe from automation, but because the thinking around the making is getting more valuable, not less.
The student who can define a problem that matters, navigate constraint, and evaluate their own work with reason — that student is on the right side of whatever comes next.
Which means some things probably have to shift.
Productive failure matters more, not less
Adults completing student work to “help” is even more damaging when the finished object is the cheapest thing in the chain. The annotation, the evaluation, the reasoning trail — that’s the expensive part now. That’s what the robots can’t do.
If you’re still weighting marks towards the finished product over the thinking that got there, it’s worth asking whether your criteria still reflect what you’re preparing students for. I’ve written about shifting the emphasis from outcome to process in this piece on rethinking how we set and assess student work.
This doesn’t mean the making stops mattering. It means the narrative around the making has to become the point.
Teaching students to see the deal
We probably also need to teach students to see deals like the one in the article with clear eyes. What does it mean to get paid for training the thing that will later compete with you. Is that a fair trade. Who decides.
That’s a D&T ethics conversation, a Citizenship conversation, and a critical AI literacy conversation rolled into one. It’s also the kind of thing no exam board has written a scheme of work for yet.
This matters because these students won’t just encounter AI in a chatbot. They’ll encounter it in an offer. Record your kitchen for $80. Let us film your workshop for training data. Wear this bodysuit so a robot can learn your trade. And they’ll need to be able to weigh what they’re giving away against what they’re getting back.
The workload angle teachers don’t talk about enough
There’s a parallel conversation here that I think is worth naming.
If the skills AI cannot replace are the ones that require human judgement, that has implications for teachers too. The work that matters most — reading a student’s reasoning, giving feedback that actually shifts their thinking, designing tasks that require genuine decision-making — is also the work that gets squeezed when admin piles up.
Protecting your time for the teaching that counts is part of the same argument. You can’t model deep thinking for students when you’re drowning in marking and planning admin. I’ve written more about how to protect that time here.
The skills AI cannot replace are becoming the ones we undervalue most in assessment.
I don’t have this fully worked out. I’m wrestling with it. But I think this article is a quiet signal to stop waiting for the DfE to tell us what to do about AI in schools. The centre of gravity in the job market is moving faster than policy can keep up, and the students in our rooms right now will graduate into whatever this becomes.
The content can mostly stay. The marking criteria probably can’t.
A practical place to start
If you want to act on this rather than just sit with the worry, I’ve put together a free guide: What If You Only Changed One Thing This Term?
It’s designed to help you update your approach to assessment without overhauling your whole practice. Six sections, one page each. Three reflective questions, a reframe, and one small thing to try. Subject-agnostic, so it works whether you’re in the workshop, the science lab, or an English classroom.
What are you changing?
I’d rather think this through with you than pretend I know.
What are you already shifting, if anything? How you teach, what you mark, what you count as a good piece of work? Or does this feel like noise that will settle?
If you want company while you work it out, PPA Buddy’s community is free to join. We’re asking the same questions you are.