# Proof Narrative: AI will replace over 50% of white-collar jobs by 2035

## Verdict

**Verdict: DISPROVED**

The claim that AI will eliminate a majority of white-collar jobs within a decade is not supported by evidence — and is actively contradicted by the most current labor market data available.

## What was claimed?

The claim is that artificial intelligence will permanently eliminate more than half of all professional, managerial, technical, and administrative jobs by 2035. This isn't a fringe prediction — versions of it have been voiced by prominent tech executives and circulate widely in media coverage of AI. For many workers, it raises a serious question: is my career at risk of disappearing entirely within the next decade?

## What did we find?

Four independent institutions — a university research lab, an AI company's own peer-reviewed research team, a major investment bank, and a leading business journal — all examined the same question and reached the same conclusion: the labor market shows no sign of the mass elimination this claim requires.

The Yale Budget Lab released a comprehensive analysis of AI's impact on the U.S. labor market in early 2026, three-plus years after ChatGPT's public launch. Their finding: the picture "largely reflects stability, not major disruption at an economy-wide level." That's not a preliminary result — it's based on years of actual employment data from a period when AI tools were already widely deployed.

Perhaps the most striking finding came from Anthropic's own research team. In a peer-reviewed study published in January 2026, Anthropic's researchers found "no systematic increase in unemployment for highly exposed workers since late 2022." This matters for a specific reason: Anthropic's CEO, Dario Amodei, issued one of the most widely-cited warnings about AI job displacement in May 2025 — yet his own company's researchers, examining actual labor market data, found no evidence of the displacement he warned about. The research also found that over half of AI work interactions involve humans and AI working together, not AI working instead of humans.

J.P. Morgan's research team found "little association between various measures of AI intensity and job growth outside of selected tech industries" — meaning that even in sectors where AI adoption is highest, the predicted mass displacement hasn't materialized. Harvard Business Review's synthesis of post-ChatGPT labor data found job postings for AI-enhanced roles actually grew, while automation-prone repetitive roles declined — a transformation pattern, not a 50%-plus elimination.

The strongest arguments for this claim don't survive scrutiny. Amodei's warning applies only to entry-level roles — a subset of white-collar work, not all of it. McKinsey's widely-cited figure that 57% of work hours are "theoretically automatable" measures what's technically possible, not what actually happens: automation of some tasks in a job typically changes that job rather than eliminating it. The IMF's finding that 60% of high-income-country jobs are "exposed" to AI similarly conflates exposure with replacement. No peer-reviewed economics study projects 50%-plus white-collar job replacement by 2035 — not even close.

## What should you keep in mind?

This verdict applies specifically to the claim as stated: permanent elimination of a strict majority of all white-collar jobs by 2035. The evidence does not say AI has no impact on work — it says the impact so far looks more like transformation than elimination. Some roles are shrinking; others are growing. Workers in highly routine, automatable positions face genuine risk. The 2035 deadline hasn't arrived, and the pace of AI development could accelerate.

The sources here also have limitations: institutional forecasts are uncertain by nature, and the credibility scoring system flagged all four source domains as "unclassified" — though this reflects a gap in the automated scoring tool, not the actual authority of Fortune, Anthropic, J.P. Morgan, or Harvard Business Review. The Goldman Sachs estimate of 6–7% net displacement if AI is fully deployed is still a real number of people whose careers will be disrupted.

The gap between what AI executives predict in speeches and what AI researchers measure in data is itself worth noting — and it's wide.

## How was this verified?

This claim was verified by searching for authoritative institutional sources that directly address AI's measured impact on employment, then checking whether those sources contradict the specific threshold of 50%-plus permanent job replacement by 2035. The strongest pro-claim arguments were tested head-on, including high-profile executive predictions and widely-cited automation statistics. Full details are in [the structured proof report](proof.md) and [the full verification audit](proof_audit.md), and the verification logic can be inspected or rerun via [re-run the proof yourself](proof.py).