For years, the conventional wisdom has been simple and disturbing: The more jobs are exposed to AI, the more likely they are to disappear. The data seemed to bear that out. But something changed in early 2025; AI recruitment trends This is the story of how the jobs most disrupted by artificial intelligence are making the most rapid comeback.
Important points
- Since the launch of Claude Code in late February 2025, the number of software development jobs in the U.S. has increased by approximately 15%, while the number of overall U.S. job openings has decreased by 7% over the same period.
- From 2022 to 2026, the occupations most exposed to AI experienced the largest declines in job openings, but these same occupations are leading the recovery from 2025 onwards.
- 71% of the growth in software jobs from May 2025 to May 2026 was for senior positions. 37% come from jobs that mention AI in the title.
- AI-enabled roles in the U.S. increased from 264 (2.6% of roles) in 2022 to 822 (8.3%) in Q1 2026, with 63% of those roles in non-tech roles.
- This pattern suggests a structural reversal. In other words, agent AI may shift the role of AI exposure from net job destruction to net job creation.
AI will have a wide-ranging impact on recruitment from 2022 to 2026
The story begins at the peak of the post-pandemic labor market boom in May 2022. From there, something predictable and troubling began to unfold. According to analysis by Guillermo Gallacher, economist at Indeed Hiring Lab. Job openings in occupations most exposed to AI-driven change decreased the most from 2022 to 2026.. Software development was hit the hardest. This decline was not a random shrinkage and tracked AI exposure closely enough to be statistically significant.
What makes this particularly impressive is that the decline started before ChatGPT was launched in late 2022. Before generative AI became a household term, early AI adoption was already reshaping employer demand. Companies weren’t waiting for a cultural moment. They were quietly recalibrating their hiring as the prospects for automation became clearer.
A rebound that no one could have predicted
Then a reversal occurred. From 2025 onwards, the occupations most exposed to AI will generally see the strongest recovery in job openings. This is a mirror image of what has happened over the past three years. The number of U.S. job postings on Indeed has increased over the year, especially when it comes to software development. almost 15% This is despite the fact that after the Claude Code began in late February 2025, the number of job openings across the U.S. fell by 7% over the same period. It’s hard to explain away the gap between software and the broader market as noise.
Mr. Gallasher’s argument is direct. “The relationship between AI exposure and job postings appears to be reversing from job destruction to job creation.”
It’s worth knowing exactly what that means. This is not to say that AI is creating jobs across the board. This is an empirical observation that the negative correlation between exposure to AI and job postings reversed over a short period of time, and rapidly.
Claude Code, Vibecoding, and Agentic AI Shift
This timing coincides with the emergence of agent AI tools that can perform complex multi-step coding tasks from plain language instructions. claude codewas launched in late February 2025 and became one of the most visible examples. In the same month, the term “vibe coding” was also coined. This is the term used to describe how AI handles the technical execution of the code while human developers focus on the vision and refinement of the product. This concept is more than just a branding exercise, it reflects a real change in the way software is built.
Whether the Claude Code sparked a rebound in software employment, or whether the two events simply coincided with a broader structural inflection point, the correlation is too clear to ignore. While there were clearly multiple factors at play in the market simultaneously, the direction of change was limited to roles exposed to AI rather than the broader labor market, indicating something more than mere coincidence.
Senior roles and AI titles drive numbers
Rebounds are not evenly distributed and the differences are very important. 71% of the growth in software development jobs from May 2025 to May 2026 was for senior roles. Entry-level positions are not leading this recovery. Employers appear to be looking for experienced professionals who can lead, oversee, and strategically deploy AI tools, rather than new graduates learning to code from scratch.
The second aspect of growth is similarly telling. 37% of the increase in software job openings was for roles that mentioned AI in the job title. These are not typical software jobs where AI is part of the job description. These are roles defined, at least in part, by AI competencies as core requirements.
The concentration of senior positions and the share of AI titles paints a consistent picture. Rather than the widespread return of the software workforce that existed in 2021, there is a growing demand for workers who can operate at the intersection of deep domain expertise and AI fluency.
Demand for AI extends beyond technology
One of the most important findings from Indeed Hiring Lab’s parallel analysis, led by EMEA Senior Director of Economic Research Pavel Adjan, is the extent to which AI demand extends beyond software. The number of US positions classified as “AI-touched” (defined as positions that mention AI in at least five posts in a given quarter) is 264 titles (2.6% of tracked titles) in 2022 and 822 titles (8.3%) in Q1 2026.
The important thing is 63% of AI-impacted roles were in non-tech occupations. Management, marketing, education and coaching are all increasing their proportions. Roles such as “Physiotherapist (AI Documentation)” and “AI Project Engineer” are emerging across sectors that would have seemed unrelated to AI just three years ago. Adjan drew a direct comparison of how computer literacy is gradually becoming a basic expectation in nearly every profession. Although this transition took decades, it ultimately universally reshaped hiring standards.
“One notable pattern is that many of the roles that have AI in the title are jobs that have been around for decades,” Adjan noted. “Employers are not only hiring AI specialists, but also adding AI to jobs that require the use of AI tools.”
Augmentation, not replacement — at least for now.
Adjan’s interpretation of AI title data is cautious but important. When a job listing includes AI in its title, data suggests it signals employer demand for enhancements rather than a harbinger of replacement. “This definitely seems to capture employers who want to incorporate AI skills into their jobs, and it’s akin to an extension,” he said. Frameworks are important for employees making career decisions. AI fluency in these situations means applying the technology to existing domain expertise, rather than pivoting to computer science.
However, this warning is real. “It is clear that if AI capabilities continue to be expected in more jobs and occupations, there is a risk that some people will not be able to get trained or get used to these tools as quickly as others,” Adjian acknowledged. The gap between workers who adapt quickly and those who don’t can grow faster than organized training systems can accommodate.
Geographic patterns and the advantages of English
The software rebound isn’t purely a US story, but geography plays a role. In most developed countries, with the exception of Germany and France, software development is increasing as a percentage of total job openings. English-speaking countries show the most consistent positive trendsAt Hiring Lab, we believe this is partially due to early adoption of agent AI tools. Many of the world’s leading AI and technology hubs are based in English-speaking countries, and these markets tend to have higher overall AI tool usage than non-English speaking markets.
This geographic bias is worth noting. If the current recovery in AI-exposed jobs is partially driven by the accessibility of English-language AI interfaces and the concentration of AI investments in English-speaking economies, it may take longer for this trend to materialize or manifest differently in labor markets in continental Europe and Asia.
Is it a structural reversal or a temporary fix?
An analytical question posed by the data, which remains unanswered, is whether this represents a true structural change or a cyclical adjustment that fades out. The mechanism behind the structural changes is as follows. Agent-based AI tools greatly increase the productivity of skilled software professionals, increasing the demand for these professionals even as the sector’s per capita output increases. This is a classic technology-driven complementarity dynamic, the same phenomenon that saw spreadsheet software increase the demand for accountants rather than eliminate it.
The concentration of growth in senior roles and AI-specific positions is consistent with that mechanism. But it also means that structural changes, if real, are currently benefiting a relatively limited workforce. The broader implications that exposure to AI will ultimately have a net positive effect on employment across all occupations remains a hypothesis supported by early data and not a confirmed finding.
What has been confirmed is that the relationship between AI and hiring is no longer unidirectional. The slope has changed. Whether this slope continues to rise, flatten, or reverse again will determine the shape of the labor market for the rest of the decade. And it’s the most important variable for job seekers, employers, and policy makers to track right now.
FAQ
How has exposure to AI affected job trends from 2022 to 2026?
From 2022 to 2026, job openings with greater AI involvement generally experienced the largest decline. However, since early 2025, the same AI-exposed occupations, including software development, have led the recovery in job openings, reversing previous trends.
What role did Claude Code play in your software development recruitment?
Since Claude Code launched in February 2025, the number of US software development jobs on Indeed has increased by almost 15%. This is in sharp contrast to the 7% decline in job openings across the U.S. over the same period. This timing coincides with the widespread adoption of agent AI tools and the emergence of vibecoding as a new development paradigm.
What software is driving the recent increase in job openings?
Growth was concentrated at the top, with 71% of the growth in software development jobs from May 2025 to May 2026 coming from senior roles. Additionally, 37% of the growth came from positions that mentioned AI in the job title, demonstrating the demand for experienced professionals with proven AI fluency.
Will the recovery in jobs exposed to AI be limited to the US?
No. With the exception of Germany and France, most developed countries have seen an increase in the share of software development jobs. English-speaking countries have shown the most consistently positive trends, likely reflecting the early adoption of agent-based AI tools and the concentration of AI investments in these markets.
Articles are created with the help of artificial intelligence and reviewed by our editorial team.
