I had a video call recently with a young woman studying communications design in New York. She’s smart, curious, already thinking about the world in ways most of her peers aren’t. Somewhere in the middle of our conversation, she used a phrase that has stayed with me since.
She called it the “sinking ship feeling.”
That’s what it’s like, she said, to be entering the workforce right now. Not panic exactly. More like a slow, creeping sense that the ground you were promised is dissolving beneath your feet before you’ve even had a chance to stand on it.
I walked away from that call thinking: someone needs to say something honest about this.
Let’s start with the data, because the data is stark.
Bank of Korea research tracking the three years since ChatGPT’s launch found that youth employment among Koreans aged 15 to 29 declined by 211,000 jobs. Of that total decline, 98.6% occurred in industries with high AI exposure. In the same period, employment among workers in their 50s increased by 209,000 in those exact same sectors. Youth job losses reached 23.8% in information services, 20.4% in publishing, and 11.2% in computer programming. The technology sector is feeling the first waves of change that are heading everywhere else.
The picture looks similar across the region. Jobstreet data published by Channel NewsAsia shows entry-level postings in Singapore fell by more than 25% in the first half of 2025 compared to the same period a year earlier, even as total job openings rose 4%. Entry-level sales roles fell 61%. Customer service positions fell 45%. Four in five Singapore employers surveyed by Remote said they had reduced entry-level hires because of AI.
Aneesh Raman, LinkedIn’s chief economic opportunity officer, compared this moment to the collapse of manufacturing in the 1980s. His phrase stuck with me: “Breaking first is the bottom rung of the career ladder.”
That bottom rung is where most careers begin. It’s where you earn the right to make mistakes on someone else’s budget. It’s where you absorb the tacit knowledge, the unwritten rules, the texture of how things actually get done, that no university can teach you. And right now, AI is eating it.
Here’s why. The tasks that entry-level workers have historically performed, research, drafting, coding, analysis, formatting, first-pass anything, are precisely the tasks that are easiest to codify. And codified tasks are exactly what AI does well. The Korea data makes this point with clinical precision: AI can substitute for a new graduate who has book-learning but no experience, while simultaneously complementing a senior professional whose value lies in judgement accumulated over years. The career ladder isn’t just getting harder to climb. For many, the bottom rungs are being sawn off entirely.
The young woman I spoke with told me her friends in computer science had started dropping out. “There’s just nothing left here,” they were saying. And she wasn’t wrong. The reason AI automated coding first is instructive. Once AI mastered code, it could use code to master everything else faster. Now that coding is largely “solved,” the same logic is rolling through marketing, legal, accounting, and the full spectrum of white-collar work.
In Hong Kong, 82% of graduates now believe AI and automation will make it harder to secure their desired job, up from 66% the year before. Graduate job openings are at a five-year low. Meanwhile, China is preparing to absorb 12.7 million university graduates in 2026, the largest cohort on record, into a labour market where youth unemployment already sits at 16.3%. Across Asia as a whole, youth unemployment rates run two to three times higher than headline averages.
The young woman I spoke with mentioned something her professor said that I haven’t been able to stop thinking about.
He told his class they could no longer be trained on hard skills, because no one knows which hard skills will still be relevant by the time they graduate. What he could give them instead was process: how to look for the right things, how to be adaptable, how to see the world through what he called “the eyes of a graphic designer.”
That framing is more sophisticated than most corporate L&D programmes I’ve encountered. He’s right. The half-life of technical skills is shrinking fast. What doesn’t shrink is the capacity to look at a problem freshly, to see what others miss, to make connections across disciplines, and to exercise judgement.
This brings me to what I believe is the most important insight in this entire conversation about AI and work, and it is not talked about nearly enough.
AI can do almost everything. But it cannot have taste.
The New York Times ran a piece in March 2026 on Silicon Valley’s scramble to understand taste as the last human competitive advantage. Greg Brockman, president of OpenAI, posted simply: “Taste is a new core skill.” Brand consultant Emily Segal put it more precisely: “Generic taste is inherently bad taste because taste is inherently relational and relative. You can’t simply replicate a superficial notion of good taste and claim it as genuine.”
Taste is the capacity to know what is right for this moment, this audience, this context, shaped by everything you have lived and read and observed. Sociologist Pierre Bourdieu called it habitus: the entire accumulated weight of your social experience expressing itself as instinctive judgement. AI can follow taste. It cannot lead it, because taste changes with humans, and AI can only ever be downstream of us.
If you are a young person trying to figure out where to place your bets, this is where I would start. Develop the capacity to make calls that cannot be reduced to a formula.
I am not going to pretend this is easy.
You are entering a labour market that was built for a different era, at a moment of profound technological rupture, compounded by geopolitical instability and a trust deficit in institutions that is entirely warranted. The sinking ship feeling is not irrational. It is an accurate reading of the situation.
But here is what I also know, after decades of working with senior leaders through crises that felt unsurvivable: the people who come out the other side are rarely the ones who had the most stability going in. They are the ones who stayed curious when everything was uncertain, who built relationships across geography and discipline, who remained genuinely useful to the people around them, and who refused to wait for permission to start.
McKinsey estimates up to 30% of hours currently worked could be automated by 2030. Most jobs will evolve rather than disappear. But the evolution will not be passive. You will have to actively remake yourself, probably more than once, possibly in ways you cannot currently anticipate. The professor who talked about giving his students “the eyes of a graphic designer” understood something important: the goal is not to prepare you for a specific job. It is to make you someone who can see clearly when the landscape shifts.
Get real experience now, even if it is messy and underpaid and not quite what you imagined. Develop taste ruthlessly, across domains, not just your own field. Go where the change is happening. Start building something of your own, even something small, even before you feel ready. And find people, whether inside your organisation or outside it, who are thinking seriously about what the world is becoming.
We need to help your generation. I genuinely believe that. And part of helping is being honest with you, rather than managing your anxiety with comfortable fictions.
The ship may be listing. But the people I respect most are the ones who learned to sail in rough weather.