The AI job apocalypse is a fantasy, says a16z. But Poland?
David George of a16z published a piece on May 6 arguing that the AI job apocalypse is a complete fantasy. He leans on a fresh batch of data: NBER 34984, Atlanta Fed (over 90% of firms reporting no employment impact in three years), Census CES 26-25 (only 5% of AI-using firms reported any headcount change, split roughly evenly between increases and decreases), and the Yale Budget Lab from April 2026. The conclusion: no apocalypse, just a slow stability with light task reallocation inside firms.
This is a direct counter to what I have been writing on this blog over the past several months. In April I covered the Great Reorg, the broader pattern of companies rebuilding around AI, and earlier I was sounding alarms about a white-collar slowdown. George argues that this is panic dressed up in the new clothing of the lump-of-labor fallacy. I’m putting his arguments next to what I see at Element‘s clients and in Polish data.
The three pillars of George's argument
The first pillar is macro data. The net effect of AI adoption on firm-level employment is currently close to zero. Roughly half of the cases that did move headcount moved it up. On U.S. earnings calls, the ratio of “AI augmentation” to “AI substitution” mentions runs 8 to 1. In other words, the data does not point to AI cutting jobs at the firm level.
The second pillar is history. The mechanization of agriculture cut the labor share of farming from 33% to 2% across the twentieth century, and displaced workers refilled factories, offices, hospitals, and eventually software. Electrification between 1900 and 1930 didn’t gut employment. It doubled productivity growth and seeded entire new consumer industries. VisiCalc and Excel didn’t kill bookkeepers either. They expanded finance teams from a million bookkeepers to one and a half million analysts and built the entire FP&A function on top.
The third pillar is specific sectors today. Travel agents lost half their roles since 2000, but the ones who stayed earn 99% of the average private-sector wage instead of 87%. Demand for software engineers started inflecting upward in early 2025, and Product Manager openings are at their highest level since 2022.
What matches George's observations
Two days ago I published a recap of the project where I built the Element-LinkedIn integration myself in vibe coding mode with Claude Code. My role on that project was not the role of a developer, because I am not one. I configured the architecture, validated technical decisions, and steered the language model toward the right calls. That is exactly the move George describes: human work pulled up the stack, higher-order skills demanded from the same person, production code emerging in the background.
I see the same pattern at my clients on the operational level (the companies that use Element and the firms where I implement AI). HR departments are not cutting recruiter headcount. They are rewriting the recruiter role into someone who designs the process, validates AI output, and runs strategic conversations. Legal departments are not letting lawyers go. They are raising the bar for what a lawyer needs to understand about technology. Operations teams automate document workflows and the freed-up time goes into analysis nobody had time for before. That is exactly the reallocation George describes in numbers.
What doesn't match
The first problem is work intensification. In February I wrote that AI in many companies doesn’t reduce work, it densifies it. The number of jobs may stay flat, as the Atlanta Fed shows, but that doesn’t mean people are doing the same volume of work as a year ago. Watching the discussions on X and LinkedIn, I see the prevailing opinion that thanks to AI we do more, not less, and free time isn’t growing.
The second problem is junior roles. George mentions it in one sentence and moves on. In practice, getting into a white-collar career in many roles is harder today than two years ago, because the routine work that used to fill a junior’s first months is closed by a language model in fifteen minutes. The Polish dev market shows it cleanly. Demand for seniors goes up, demand for juniors shrinks. Net employment stabilizes, but the entry path narrows and the first three years of a career look nothing like the path of someone five years older.
The third problem is the period the data comes from. George’s argument is built on data from 2024 to 2026, when AI adoption inside firms is still shallow. Most of the deployments I have seen up close in the past twelve months stopped at internal chatbots or single-document automation. The real wave of AI agents working inside corporate systems is just starting. Concluding “no effect” from data taken before the wave is the equivalent of judging the impact of electrification in 1905.
Polish data doesn't quite fit U.S. macro
George’s argument rests on U.S. macro data. Poland differs in two ways. First, we had seven straight months of declining job postings in 2025, which signals a distribution effect that the Atlanta Fed numbers don’t catch. Second, Poland has a structurally larger share of BPO, SSC, and pure operational roles, which are the direct target of AI agents. Poland will not necessarily follow the same path as the U.S., where the sector mix is different.
That doesn’t mean the apocalypse is materializing. It just means George’s stability conclusion has to be imported carefully. Element ATS data on Polish job postings shows the change is already underway, just on a different curve than in the U.S., because the market structure is different.
What this means for CEOs and HR
George is right that the “mass unemployment from AI” scenario is currently a fantasy that the data doesn’t support. That matters, so the narrative doesn’t tilt into panic, because panic produces bad budget decisions inside companies: hiring freezes, blocked promotions, training cuts. All on the assumption that nobody will be needed in a year. That’s a self-inflicted wound.
If you are running a company, your job in 2026 and 2027 is not to cut headcount, it is to redesign the role descriptions for real work with AI. You have to accept that the same person on the same paycheck now has a wider scope, and that a junior gets a different starter task than the previous cohort. Without that, the employment data won’t move and the operational stress inside the company will be enormous.
DISCOVER ELEMENT!
Maciej Michalewski
CEO @ Element. Recruitment Automation Software
Recent posts:
The AI job apocalypse is a fantasy, says a16z. But Poland?
David George of a16z says the AI job apocalypse is a fantasy. I test his thesis against Polish data and a fresh vibe coded experiment of my own.
Vibe Coded Element x LinkedIn integration is live: project recap
LinkedIn approved Element’s public XML feed, fully built in vibe coding mode with Claude Code. What that means for people who don’t write code.
danehr.pl: a new data hub for the Polish labour market
Polskie Forum HR has launched danehr.pl, a portal that gathers employment, wages, unemployment and labour market reports in one place. Element is the technology partner of PFHR and feeds one of the portal’s sections.
The Great Reorg: companies are rebuilding around AI
Foundation Capital interviewed 25 companies and found that AI is not just speeding up work — it is forcing organizations to rebuild from the ground up around four human roles.
Gallup 2026: the workplace is cracking and managers lose drive faster than the rest
Gallup State of the Global Workplace 2026: engagement fell to 20%, managers lose drive faster than their teams, and the AI paradox is clear: personal productivity rises, organizational does not.
A Polish-founded robot that drives screws faster than humans
A Polish-founded robot that drives screws faster than humans When people think about automation, they usually picture chatbots, image generators, and coding assistants. Meanwhile, AI
