B EN Unknown·Report·

AI and Job Postings: From Destruction to Creation?

B · Useful, limited Short Indeed Hiring Lab analysis by Guillermo Gallacher using Indeed's own postings data; credible platform data and a notable reversal, but uncontrolled correlations in a blog-length note.

Summary based on the full text·Original: about 6 minutes to read

In short

Indeed Hiring Lab, 8 July 2026. Software development postings in the US have rebounded since early 2025 while overall postings kept falling. Between 2022 and 2026 the most AI-exposed occupations lost the most postings, but over the past year the most exposed occupations have rebounded the most. The rebound is concentrated in senior and AI-titled roles.

Key numbers

  • almost 15% vs -7%change in US software development postings versus all postings since Claude Code's launch in late February 2025
  • 27.5%how far software development postings remain below their pre-pandemic level
  • 71%share of the May 2025 to May 2026 rise in software postings coming from senior roles
  • 37%share of that rise from jobs with AI in the title

As reported by the source.

Read the original on hiringlab.org

English·Web page·opens hiringlab.org

What it reports

  • Since 24 February 2025, US software development postings on Indeed rose almost 15% while all postings fell 7%.
  • Software postings are still about 27.5% below February 2020; overall postings are roughly at their pre-pandemic level.
  • The software share of postings is rising in most large developed economies, with Germany and France the exceptions; English-speaking countries show the most consistent rise.
  • May 2022 to May 2026: more AI-exposed occupations declined more, a statistically significant uncontrolled relationship.
  • May 2025 to May 2026: the relationship flips; more exposed occupations rebounded more.
  • 71% of the software rebound is senior roles and 37% is AI-titled roles, with overlap, consistent with seniority-biased change.

How they got it

Seasonally adjusted Indeed postings indexed to chosen dates, cross-country posting shares, and scatter plots of occupation posting changes against a generative AI exposure measure.

Limits

  • The author repeats that correlation does not imply causation; no controls for interest rates or the tech cycle.
  • The Claude Code date is an anchor, not an identified cause.
  • Postings are not hires; one platform's data.

Why it matters for companies

Demand is returning for experienced engineers who work with AI, not for juniors. Employers competing for senior AI-fluent talent face a tighter market than the overall slowdown suggests, while entry-level pipelines keep shrinking.

Summary based on the full text. Read the original for the full argument and data.Read the original on hiringlab.org

Tags

Same week

All of Week 28