For decades, new graduates could count on the same trajectory: get a degree, land an entry-level job at a good firm, and climb the ladder. That ladder is now missing its bottom rungs.
Since 2019, the share of entry-level roles at major tech firms has halved. Employment rates for 20- to 24-year-olds are slipping, internships are harder to find in every sector, and companies moving towards building smaller, more experienced teams — with generative AI as their silent partner.
This shift is not just about efficiency; it’s about composition. When firms skip hiring younger workers — typically lower paid — the result is lower overall employment rates for recent graduates, even as wage averages rise for older, more experienced employees. The message is clear: AI is squeezing the middle and eliminating the low end of the skills spectrum.
In this new market, “mediocre expertise” is being automated away. What remains valuable is specialised, adaptive expertise, the kind that can be amplified by generative AI, not replaced by it. A handful of AI model engineers are getting pro athlete-level offers, as seen in recent news, but for everyone else, survival will depend on becoming fluent in AI-augmented thinking.
GenAI literacy is not about knowing which prompt to type into ChatGPT. It’s about integrating large language models into your workflows in order to better test hypotheses, synthesise research, challenge assumptions, and generate creative solutions. It’s the cognitive equivalent of having an exoskeleton: you still do the thinking, but your reach, stamina, and speed explode.
Without that literacy, AI doesn’t just replace entry-level tasks — it removes the very training ground where skills are developed. Graduates risk a career of running behind the technology instead of running with it.
The solution is simple in principle, but urgent in practice: make GenAI literacy as fundamental in education as reading, maths, and critical thinking. Start early, update constantly, and teach AI not as a magic trick but as a thinking partner.
We’ve seen what targeted cognitive training can do, from working memory exercises in childhood that dramatically improve long-term outcomes, to specialised expertise that commands premium pay in this new AI era. The next generation’s edge will come from learning how to think with AI, not letting AI think for them.
In the new job market that is emerging, the winners will not be those who can code or those who can “beat the machine”. The real winners will be those who know how to make the machine work for their mind.
Let’s build human-centred futures—together.
Thanks for reading.
— HM