If you happen to live with a teenager, or you just know one and pity the adults who share their space, there’s a good chance your conversations revolve around revision timetables, exam stress, and occasional, fraught, conversations about “what happens next”. GCSEs and A-levels have a remarkable ability to turn entire households into low-level anxiety management operations – as if the impact of all those hormones wasn’t enough already.
At the same time, the adults (me) attempting to reassure angsty exam candidates are scrolling through headlines about AI replacing jobs, graduate schemes shrinking, and something the Financial Times recently described (helpful, thanks…) as an AI “jobpocalypse”. *
Timing-wise, it’s not ideal when you’re on a ‘let’s keep everyone calm and focussed’ mission.
The FT piece pulled together a growing body of evidence suggesting that entry-level white-collar work may be among the first areas genuinely reshaped by generative AI. Which comes as a surprise to precisely none of you. White collar jobs won’t necessarily be wiped out overnight (hiya!), despite the more dramatic takes online, but most will be changed enough to create real uncertainty for people trying to work out what to study, where to train, or whether university still guarantees the sort of career progression previous generations expected.
And if we’re honest, the difficult bit is that the adults offering advice don’t know either. Even the ones that sound really confident.
Our lack of confidence in our own careers advice feels new
For years, careers advice has largely worked on the assumption that the world changes steadily enough for sensible recommendations to stay sensible. The opening of Trainspotting resonated for a reason: “Choose life. Choose a job. Choose a career…”
Now, I’m not sure anyone can confidently point to a truly “safe” category of work in the way they once could (although when my neighbour said her son was studying how to tan deer hides by hand, I did think that might be fairly AI-proof). That doesn’t mean there are no good jobs, or that young people are doomed. It just means the old certainty has gone a bit wobbly.
The awkward irony is that some of the careers traditionally presented as secure and aspirational for the most academically successful students now appear to be the most exposed to AI disruption. Research from King’s College London found firms highly exposed to AI reduced junior hiring by nearly 6%, with job postings in highly exposed occupations dropping by over 23%. Graduate recruitment firms are reporting employers planning to hire fewer graduates specifically because of AI. Meanwhile, entry-level roles in areas like programming have fallen sharply among younger workers.
Again: not extinction. Change. Don’t panic.
And the public conversation about this often swings wildly between two equally unhelpful extremes. On one side, the “learn to code” crowd, who spent years treating software engineering as the universal answer to future-proof employment. On the other, the full robot-apocalypse brigade who appear to believe no human will ever work again by next Thursday.
Reality is, as ever, messier than the headlines
One of the more interesting patterns emerging from labour market data is that exposure to AI does not always equal replacement. In some sectors, people using AI effectively become more valuable rather than less (I think a lot of you know this already). Cloud engineers, analysts and finance professionals are still needed – but the nature of the work is changing. Fewer repetitive junior tasks; more oversight. More judgement. More responsibility for problem-solving and checking whether the machine has just confidently invented nonsense.
That last part may turn out to be one of the defining skills of the next decade: knowing when not to trust the machine, and being able to explain what’s happening.
Which is why some of the more comforting commentary about “human skills” also feels slightly incomplete. We keep hearing that creativity, empathy and communication will save us all. Maybe. But generative AI is already reasonably good at mimicking large chunks of those things too. If your definition of creativity is “produce a passable first draft” or your definition of communication is “write a competent email”, the machines are already way ahead of you.
The more resilient human skills will likely be things like judgement, adaptability, resilience, context, ethics, trust, leadership and the ability to navigate ambiguity without collapsing emotionally over each tiny challenge. Which can all sound like a stretch when you’re watching a teenager have a meltdown because you said they couldn’t go to a gig way across town the night before their main maths exam (and if you’re thinking that’s a very specific example: correct).
The other problem is that our education systems are still largely built around a model of individual production and knowledge recall, while many workplaces are rapidly evolving towards something more collaborative and tool-assisted. We still assess students as though the highest-value skill is producing work entirely alone under timed conditions, even though most professional environments increasingly involve working alongside software, systems, OTHER PEOPLE and AI tools.
That doesn’t mean exams are pointless. It does mean the relationship between education and employment is becoming less straightforward. I’d also gently suggest that the shift schools in England made under Michael Gove — moving away from coursework and towards more rote learning and timed exams — may not have aged especially well for the world we’re now entering.
There’s also an emotional layer we’re only just starting to understand: many parents (me) are trying to advise children for a labour market we don’t really recognise. Entire generations grew up believing that if you got the grades, chose carefully, and built expertise in a respectable field, stability would broadly follow. Try landing that when they’re friends with a successful YouTuber and they already know they’ll never be able to afford a house in London.
But there is a more optimistic way to look at this.
Historically, technological shifts do not just eliminate work; they change it. Around 70% of UK workers are in jobs containing tasks AI could potentially perform or enhance, but that is not the same thing as saying 70% of jobs disappear. The challenge is that transitions are awkward, especially for younger people trying to enter the labour market at the point the change is happening, often following courses of study designed before the shift really hit.
So what do we actually tell our kids?
Honestly? Probably something a bit less specific than previous generations got away with, although Doctors, Lawyers and Accountants will probably still do okay, because people with the right certificates will always be on the hook for decisions.
The most useful careers advice now may be less about identifying one magically “safe” profession and more about helping young people become adaptable enough to cope with change. Which probably means encouraging them to:
- Get genuinely good at something – aptitude and depth matters.
- Keep learning throughout your life, because the tools and industries will keep changing. Learn how technology works rather than fearing it.
- Understand how to work with AI, but also understand its limitations and know when not to trust it.
- Build confidence in talking to people, working in teams and explaining complicated things well.
- Practise solving messy problems where there isn’t one obvious right answer.
- Stay open to unexpected opportunities and entirely new kinds of jobs that do not exist yet.
- Avoid writing off subjects too early, because combinations of skills may become more valuable than specialisation.
- Accept that the path may change several times along the way – and that this is increasingly normal, not a failure.
A lot of successful adults already have careers that barely resemble the thing they originally trained for, and I say that as someone who decided to major in Film Studies. The goal for young people is not finding the one career untouched by technology; it is becoming the kind of person who can keep adapting as the world changes around them.
Which, in fairness, may have always been the real point of education anyway.
PS: This is the first in a short series asking what humans are actually for when machines start getting better at more and more things. Next up will be the strange economics and unintended consequences of automation, inspired by self-checkouts and shoplifting in my local Tesco Metro. And after that, the really awkward one: empathy. Or at least the performance of empathy. Because if humans “fake” professional empathy more often than we admit, and AI can simulate it surprisingly well, we may need to think much harder about what we actually mean when we say we want more “human” interactions…
*I’d offer you a link, but it’s just annoying if you’re not a subscriber. You can take my word for it on the general gist or ask Perplexity for a summary, that’s my advice.
