AI and work: why everyone stays silent — and what we still need to say
Since AI changed my working day, one thing is clear to me: we're facing a bigger upheaval than most people suspect. And yet there's still remarkably little conversation about what it means for the world of work.
Since I started integrating AI tools into my day-to-day work, a lot has changed. With Claude Code I automate video editing, have an agent translate our company blog directly inside WordPress, and build tools I would once have needed external contractors for. Not just simple things, but more complex processes too, the kind that used to take weeks or tie up several people. Text, concepts, emails, programs: all in all, my productivity has at least doubled. At the same time I now do things that simply weren't possible before, at least not without technical help.
What has happened over the past few weeks is a different order of magnitude again. Since Claude Code started running on Opus 4.5/4.6 and OpenAI followed with Codex and ChatGPT 5.3/5.4, agentic programming has made a leap I genuinely didn't expect. The models seem to have been trained specifically to drive even the most complex tasks all the way to the finish, efficiently and on their own. Every day I'm surprised by what Claude Code builds for itself in the background to reach the goal faster or get around an obstacle.
That finally convinced me we're facing an upheaval bigger than most people suspect. It was already noticeable before. But now it's so visible that you start wondering why the rest of the world hasn't yet registered what's rolling toward it. Either way, there's still remarkably little conversation about what all of this means for the world of work as a whole.
The math no one says out loud
When someone is twice as productive and also takes on tasks that used to be spread across other shoulders, at some point a company needs fewer shoulders. Not tomorrow, and in mid-sized businesses probably not through big waves of layoffs. It happens quietly:
- A position that isn't filled again.
- A freelance contract that falls away.
- A team that comes out smaller on the next project.
For aichholzer.me/jobs I analysed the occupations of the 4.4 million people currently employed in Austria. Around 43.6%, that's 1.9 million people, work in an occupation where a high degree (level 7–10 of 10) of automation through AI is possible.
Important: this isn't about all of these people, or their jobs, being completely replaced. It's about large parts of their daily tasks becoming automatable through AI. That's exactly what I experience in myself, and it's exactly what makes the change so hard to see: no one becomes redundant overnight. But piece by piece, the same work needs fewer people.
Why the silence is systemic
The silence runs through every level. Employees stay quiet right now because transparency means either more work or fewer colleagues. According to an Ivanti study of more than 6,000 respondents, 32 percent hide their AI-driven productivity gains and 36 percent treat them as a private advantage. That won't last forever. As soon as companies recognise this efficiency systematically, the quiet productivity gain turns into an official expectation.
Companies stay quiet for similar reasons, only more expensively. When Anthropic launched Claude Cowork in early 2026, an AI agent that can read, analyse and create files and hooks into existing enterprise software via plugins, the largest SaaS companies lost around 285 billion dollars in market value in a single day. Thomson Reuters dropped 18 percent. Say out loud that AI can replace whole professional fields, and the market punishes you.
And politics simply has no popular answers.
The comfort of new jobs
I know the standard argument against worrying about an AI future very well: every technological upheaval has created new jobs. Historically that's true, but it leaves out two things. During the Industrial Revolution, real wages stagnated for decades before new industries made up for the losses, and the changes mainly affected physical labour. AI affects far more people and is moving faster than anything before it.
A Brookings analysis of historical US retraining programs shows they produced no statistically significant improvement in employment or income. Not everyone becomes an AI specialist, and certainly not everyone becomes a nurse. People just don't like to say so.
As an optimist, I'm convinced we're heading toward a wonderful destination, perhaps a world where people no longer have to work full-time and have more time for themselves. The only question is how long the road there takes, and how bumpy it gets.
Who's winning today, and where that stops
Let's be honest: anyone using AI productively today is among the winners. Less stress, more output. Leadership often hasn't even grasped what has changed. In my workshops I see both sides. The excitement of those who can suddenly do things that seemed impossible before. And the quiet fear of those who sense what it means for team sizes.
Cancer is cured, the economy grows at 10% a year, the budget is balanced — and 20% of people don’t have jobs.
So the overall balance can be positive, and millions of people can still end up without work.
So what do we do?
We don't know exactly yet. But there are approaches beyond the reflexive cry of “retraining!”. The four-day week works in more than ten countries, with stable productivity and less stress. Sam Altman's UBI study with 3,000 participants found that people receiving 1,000 dollars a month in basic income worked almost as often as those without it; they just chose their jobs more deliberately. In Switzerland, the Federal Council is examining, by the end of 2026, whether AI calls for a fundamental change to the tax system.
Most of all, what we need is honesty. The numbers are uncomfortable. But they don't get any better if we pretend they aren't there.
