I build an open-source HR platform, so people keep telling me two things: AI will soon manage everyone, and AI will soon replace everyone — so why bother. My company dies if the second group is right. So instead of arguing, I went through the research. What I found is more specific than "replacement," and in some ways worse.
Mass replacement is not in the data
Start with the strongest evidence available. In Denmark, researchers matched chatbot adoption with government payroll registries: 25,000 workers, 11 of the most AI-exposed occupations. Two years after ChatGPT launched — no measurable effect on earnings or hours. The confidence intervals rule out anything larger than 2%.
Aggregate employment in the US barely moved either. The mass layoffs everyone predicted for 2026 didn't happen.
But something else did.
The entry point collapsed
A Stanford / ADP study tracked employment by age in AI-exposed occupations. Workers aged 22–25: down 16%, while older colleagues in the same occupations held steady. For young software developers the drop is close to 20% since late 2022.
The mechanism matters: companies didn't fire juniors. They stopped hiring them. Indeed's data shows the market tilting toward seniority: postings recover, but the growth goes to senior roles.
If you're under 25 and your job search feels broken — it's not you. The bottom rungs of the ladder are gone.
Meanwhile, the productivity story is shakier than the slides say
METR ran a randomized trial with experienced open-source developers working on their own repositories. With AI tools they were 19% slower. They believed they were about 20% faster.
I use these tools daily and they're clearly useful for parts of the work. But that perception gap should make you skeptical of every "AI made us 40% more productive" slide. It also explains why real adoption is slow: Census data shows only about a fifth of US companies use AI at all. The distance between "the model can do it" and "the company actually works this way" is measured in years.
Two more things quietly broke
Trust in hiring signals. CVs are generated in a minute. Take-home tasks are solved by models. Interviews are coached in real time. Gartner expects 1 in 4 candidate profiles to be fake by 2028. Every cheap signal the industry relied on for decades is losing value at once.
Skill lifetime. Gartner projects the half-life of technical skills at 2–5 years, down from 8–12 historically. Whatever framework you're an expert in right now — that expertise has an expiry date.
My 10-year forecast
Four bets, not ten. The rest would be noise.
Teams get smaller and stranger: a few people directing dozens of agents. Not because of org-chart fashion — because accountability doesn't compile. The EU already classifies hiring and promotion decisions as high-risk, with a human legally responsible. A model can't sign a contract or answer to a regulator.
The senior shortage arrives around 2032. Every junior not hired in 2024–2027 is a senior who doesn't exist in 2032. Companies will rediscover internal training the hard way. If you're a junior who survives the broken entry, you'll be scarce. Scarcity pays.
The prompting premium fades. Wage premiums for tool skills always fade as the tool spreads — "computer skills" commanded one in the 1990s. What holds value is domain depth plus the willingness to take responsibility for output you didn't type yourself.
Proof of skill becomes the scarce resource. When CVs, interviews and take-homes are all fakeable, the question "what can this person actually do" gets expensive to answer — and worth answering well.
Where I land
That last point is what I'm building: an open-source (AGPLv3) talent platform — competencies, assessment, development — for a world where the old signals are dead. Code is on GitHub.
And the honest version of my bet: if AI truly replaces everyone, my product dies — together with every other B2B product, because nobody is left to buy software. That's the only scenario I don't plan for. In every other one, figuring out what humans can actually do gets harder and more valuable at the same time.
If you're under 25: how did your job search change this past year? And if you're senior — when did your team last hire a junior?
Top comments (0)