What's on my mind
I've been thinking about the BT team a lot this week, and I realized something about the decisions we're making at Blended Teaching. Take Braden: we hired him a couple of years back, right out of UNG's Real Estate program (thanks for the intro, James!)
Braden didn't have 'AI' skills, but he did have a lot of traits that we want in members of the team. Judgment, the ability to work under pressure & ambiguity, and a willingness to tackle something big and unknown & figure it out along the way.
With the exception of developers, everyone we are hiring, or looking to hire, has skills that look like these. Knowledge & the 'ability to use AI' matter much less to us now. So this all got me wondering: what are other employers looking for?
Teaching notes
Nobody wants a professional prompter
We need to talk about what employers actually want from our graduates now that AI is in every office, because it's not quite what the flurry of news articles over the last couple of years predicted.
So what are employers asking for?
The short answer: AI skills, yes. But mostly the stuff AI can't do.
Let's start with the bit that's no surprise. In the US, more than a third of entry-level jobs now require AI skills, nearly triple the figure six months earlier (NACE). Jobs asking for AI skills carry a 62% wage premium, according to a PwC trawl of over a billion job ads 👀 (PwC).
But look at which AI skills. When NACE asked employers what they meant, "analyse and revise AI outputs" made the shortlist alongside picking the right tool and writing decent prompts. In the UK, the ISE says AI capability is becoming a baseline expectation, meaning everyday competence rather than wizardry. The skills rising in value? Critical thinking, judgement, adaptability and communication (Wonkhe).
Now for the twist that should make every programme leader sit up. PwC found that AI-exposed entry-level roles in the US are 7X more likely to require traditionally senior skills like leadership and judgement. Those "seniorised" junior roles have grown 35% since 2019, while other entry-level roles have shrunk 10%.
Read that again. Employers want juniors who arrive with judgement. Judgement used to be what the junior job gave you. 🤔
And the job that gave it to you is harder to find. Stanford's payroll data show employment for 22- to-25-year-olds in AI-exposed jobs is now 19% below trend. The losses are concentrated in roles built on codified knowledge, the textbook, examinable kind (Stanford). Which is, awkwardly, a fairly good description of what a degree certifies.
Right, so what do we do with the curriculum?
Here are a few questions worth lobbing at your next programme meeting. Bring biscuits (cookies, for readers in the US, not those odd savory scone-things).
If AI can pass it, what is it certifying? Pick a module. Which assessments could a decent AI get through? Those tasks might still be worth teaching, but they're no longer proof of anything to an employer.
Are we teaching students to check AI, or just to use it? Try handing students a confidently wrong AI-generated analysis and marking them on how well they take it apart. PwC has already shifted its audit training towards exactly this (Fast Company).
Have we protected the foundations? This is the counter-intuitive one. A Microsoft and Carnegie Mellon study found that people scrutinise AI output when they're confident in their own knowledge, and switch off when they trust the machine (The Register). Anthropic's trial found junior developers learning with AI scored 17% lower on comprehension, and worst of all on debugging (Anthropic). You can't spot the error if you never learned the thing. So learn it by hand first, then learn to supervise the machine doing it.
Does our assessment tell learning-with-AI apart from proving-what-you-know? Two-thirds of UK employers already suspect candidates use AI to misrepresent their skills. The University of Sydney's "two-lane" model is one tidy answer: secure vivas and orals to verify, and open AI-friendly tasks to learn (Sydney).
*We like the Sydney two-lane model. We expanded on it in a recent webinar on assessment integrity. Watch it here if you haven't already.
Where do students exercise judgement on something real? It looks like firms are quietly retiring apprenticeships, & USC economist Miao Ben Zhang argues universities are best placed to rebuild it (Zhang). Think law clinics, live client projects and residency-style placements, where the output can't be produced without judgement. Are placements the spine of your degree, or a nice optional extra?
The last word
A design director quoted in Fast Company summed up the employer mood in one line: "I don't want professional prompters. I want designers."
The graduates who thrive won't be the ones who learned AI instead of their discipline. They'll be the ones who know their discipline well enough to tell when the AI is wrong, and who've practised acting on that before anyone paid them to.
Like we always say, you've got to earn the right to automate!
Worth a look
I Built AI Clones of My Coworkers. Things Got Weird
The world of work is getting strange.
Sam