Fifty-five thousand. That’s how many US jobs in 2025 were explicitly tied to AI-driven automation—12 times the figure from two years prior. Another 9,200 cuts landed in Q1 2026. Meta alone shed roughly 8,000 roles while simultaneously pouring $115–135 billion into AI infrastructure. Capital moves fast. Labor absorbs the delay.
Yet Washington’s labor data systems treat this upheaval as a minor statistical blip. A Congressional Research Service report from early June describes federal labor data infrastructure as dangerously outdated. Official systems miss the nuance: employers cite AI in layoffs, but monthly surveys lack occupational granularity, classification codes update too slowly, and state unemployment insurance data remains siloed. Policymakers are flying blind.
The Boston Consulting Group projects 50–55% of US jobs will be reshaped by AI within two to three years, with only 10–15% eliminated outright. Most roles will transform—workers will amplify output with AI tools, rebalance duties, or gain new capabilities. But the Dallas Fed found that in the 10% of sectors most exposed to AI, employment fell 1% since ChatGPT’s debut. Computer systems design shed 5% of jobs. Young workers under 25 bore the brunt; older cohorts held steadier.
Geoffrey Hinton warns 2026 could mark sharp acceleration. Federal Reserve Governor Lisa Cook notes the timing problem: AI may displace workers before generating new positions. Unemployment among recent college graduates is rising, even as the overall rate sits near 4.3%.
The CRS urges Congress to add occupational questions to monthly surveys, update job classification codes more often, expand state UI data, and invest in better statistical tools. Without these steps, all other policy responses risk missing the target.
The asymmetry favors owners of AI infrastructure. Service-sector ripple effects trail capital spending by two to four quarters. Governments that treat this as a data problem first—then a policy problem—stand the best chance of guiding the transition instead of reacting to its fallout.
Source: Webpronews