AI on the Factory Floor: Evidence from Manufacturing Job Postings
A · Strong evidence Federal Reserve Board FEDS Note by Maria Tito using Lightcast posting-level data validated against JOLTS and published skill taxonomies; method and caveats are shown, numbers are checkable.
Summary based on the full text·Original: about 8 minutes to read
In short
A FEDS Note (30 September 2026) by Maria Tito of the Federal Reserve Board. Task-based exposure measures rank manufacturing as relatively unexposed to AI, yet manufacturing job postings now ask for AI skills more often than the economy as a whole: 11% versus 8%. The rise came in the second half of 2025 and is driven by machine-learning skills; generative AI skills remain rare. Postings that ask for AI skills advertise wages about 70% above the manufacturing average, and production-worker postings have carried a gap of roughly 30% since 2023.
Key numbers
- 11 percentshare of manufacturing job postings requiring AI skills after the surge in the second half of 2025, versus 8 percent economy-wide
- around 70 percentaverage wage premium of AI-skill manufacturing postings over all manufacturing posted wages
- roughly 30 percentaverage AI wage gap for production-worker postings since 2023
- 35 percentaverage share of manufacturing postings requiring computer skills, versus 24 percent across all sectors
- under 1 percentshare of manufacturing postings asking for generative AI skills
As reported by the source.
English·Web page·opens federalreserve.gov
What it reports
- Lightcast postings track JOLTS openings closely; manufacturing is about 6.5% of new hiring demand in both.
- Computer skills appear in 35% of manufacturing postings on average, 24% across all sectors.
- AI skill requirements were flat and modest until early 2025, then rose to 11% in manufacturing and 8% overall. In 2018 Acemoglu et al. found about 1% for both.
- Machine-learning requirements rose notably from mid-2025. Generative AI skills are under 1% of postings and essentially absent from production postings through the first half of 2026.
- Production occupations are around 50% of manufacturing employment.
- AI-skill postings advertise wages around 70% above all manufacturing postings, with high volatility; the premium appeared before the volume rose.
- For production workers the AI wage gap was near zero until 2023 and has averaged roughly 30% since.
How they got it
Posting-level Lightcast data on manufacturing establishments, data through July 2026. AI skills follow the Acemoglu, Autor, Hazell and Restrepo (2022) method; computer skills follow Deming and Kahn (2018). Wage comparisons use postings that report wages, about 50% of postings in recent years.
Limits
- Postings show stated requirements, not the skills of people hired, and miss offline and internal hiring.
- Wage differentials are raw averages and may reflect differences in role, location and employer.
- The small sample of AI postings makes the premium volatile.
- US data only.
Why it matters for companies
Exposure scores based on tasks underestimate how fast physical-work sectors are adopting AI. Manufacturers are already writing machine-learning skills into job ads and paying for them, including for production roles. For workforce planning in industrial and supply-chain functions, skills frameworks and pay bands for operators and technicians need an AI dimension now, and the premium suggests that buying these skills on the market costs more than building them internally.
Summary based on the full text. Read the original for the full argument and data.Read the original on federalreserve.gov