Overview
This series explains the early U.S. labor-market response to generative AI by separating total firm job postings, coding-intensive employment, industry conditions, task allocation, and 2024–34 projections. Federal Reserve evidence through November 2025 finds no negative total-posting response, while coder employment growth slowed after 2022 and lagged industry paths. Firms can shift occupational mix or defer replacement hiring without cutting total recruiting; the evidence supports a concentrated, conditional adjustment, not a complete count of AI-caused job loss or a settled long-run outcome.
Learning outcomes
- Stable total job postings do not show that every occupation or worker is unaffected.
- Coding-intensive employment growth slowed after 2022 but continued to rise in the raw series.
- Industry weakness explains part of the coder slowdown; the remaining gap is consistent with occupational change, not proof of AI causation.
- Firms can reallocate tasks and defer replacement hiring, so coder opportunities may weaken before total recruiting falls.
- BLS 2024–34 projections show uneven growth across technical, administrative, and service occupations, not a uniform AI effect.
Keywords
Preview every chapter · Unlock full text, audio, and video from chapter 4
PlansFAQ
How can firms keep posting jobs while coder employment slows?
Firms can keep total postings stable by shifting openings across occupations, slowing coder replacement hiring, or reallocating coding tasks to other roles.
Did coder employment collapse after ChatGPT?
No. The raw coder-employment series continued to grow after November 2022, but its annual growth trend slowed by roughly three percentage points after accounting for industry paths.
What does the industry counterfactual show about AI and coder jobs?
It shows coder employment lagged the path implied by the industries employing coders, supporting an occupation-specific change while leaving other causes unresolved.
Do these findings prove how AI will affect jobs long term?
No. The studies do not identify the full causal effect or settle long-run demand; lower coding costs could reduce near-term hiring, expand software demand, or do both over time.