Public projects · Data investigation · Labour market
The Quiet Freeze
Did hiring drop more for the jobs AI can actually do?
Personal project on public data. Not client work. An independent analysis, not linked to any company mentioned.
−39%postings for the most AI-exposed jobs since ChatGPT
−21%gap reached in October 2023
36months the gap has not moved
The data
Indeed job postings for 47 occupational groups, from 2020 to October 2026, matched with the Anthropic Economic Index, which measures how much of each occupation’s work AI already does.
How it was built
Power BI, more than 160 DAX measures, scenario forecasting, and a full Arabic version alongside the English one.
What I found
- Since ChatGPT launched, postings for the most AI-exposed jobs fell 39%, against 26% for the least exposed.
- The decline began in early 2022 with higher interest rates. Low-exposure jobs recovered to 25% above pre-pandemic levels; the exposed ones are still 12% below.
- The same analysis for the UK, Germany, France, Canada and Australia points the same way in all six countries, and is statistically significant in five.
- Job ads asking for AI skills went from 2.3% to 7.2% of US postings.
How I tested it
I tried to break the finding: against the 2022 hiring boom, against trends that were already there, and without tech jobs. Most of it held. One test got weaker, and it stays in the report.
Screens
Next project
The Delivery Bill
Want a report like this on your own data?
Get in touch