AI Employment Impact: A Research-Based Analysis of 2023-2024

Research Analysis: Two Years of AI Workforce Impact

Data Sources

• Challenger, Gray & Christmas monthly layoff reports (AI as category since 2023)
• Bureau of Labor Statistics occupational projections incorporating AI
• OECD workforce automation risk assessments
• Goldman Sachs Global Research — 300M jobs globally affected estimate
• McKinsey Global Institute — 30% of US work hours automatable by 2030

Quantifying the Impact

Challenger Gray data shows approximately 24,000 AI-attributed job cuts in 2023 and roughly 12,700 in 2024 (their methodology). However, their figures are widely considered to be significant undercounts — they only capture formally announced layoffs where AI is explicitly cited.

• 2023: ~35,000 core weighted estimate
• 2024: ~110,000 core weighted estimate
• Cumulative through 2024: ~145,000

The Attribution Problem

A key methodological challenge: the gap between what companies say publicly and what they disclose formally. When New York State added an AI checkbox to WARN filings in March 2025, zero of 160 companies checked it — despite many of those same companies publicly citing AI as a restructuring driver.

This suggests either: 1. Legal teams advise against formal AI attribution to avoid liability 2. The actual AI contribution is smaller than public narratives suggest 3. Both — companies use AI as PR narrative but avoid legal commitment

Academic Findings

The St. Louis Fed found a 0.57 correlation between generative AI adoption rates and unemployment increases — statistically significant but not yet definitive proof of causation.

The Dallas Fed found employment in high AI-exposure jobs fell 13% for ages 22-25, suggesting entry-level workers bear disproportionate impact.

Brynjolfsson et al. documented 15% average productivity gains from AI assistance, with the largest gains for less-experienced workers — suggesting AI may flatten experience hierarchies.

Sector Analysis

Conclusion

The data supports a central finding: AI-attributed job displacement is real, growing, and spreading across sectors. However, the magnitude remains debated due to attribution methodology, and the net employment effect remains ambiguous as AI-created jobs partially offset losses.