Only 4% of AI-exposed job postings ask for AI skills, Mango Thrive study finds

A Mango Thrive study of 4,192 U.S. job postings across 21 occupations found a weak link between AI exposure and employer demand for AI skills. The research suggests companies in the most AI-exposed roles are still rarely listing AI skills in job ads, even as copywriters, web developers, and software developers show the highest demand.

Why it matters: - The study suggests employer demand for AI skills is still concentrated in a small slice of job postings, even in occupations most exposed to generative AI. - That gap could slow how quickly AI-related skills show up in hiring signals across the labor market. - The findings also matter for job seekers watching which roles are actually rewarding AI experience.

What happened: - Mango Thrive analyzed 4,192 U.S. job postings across 21 occupations. - The sample included 14 occupations classified as highly exposed to generative AI and 7 low-exposure control occupations. - The postings were collected between June 3 and September 1, 2026. - The study used the International Labour Organization’s 2025 generative-AI exposure index to choose the occupations.

The details: - Just 4.2% of postings in the 14 highly exposed occupations asked for AI skills. - Across all 21 occupations, 2.9% of postings asked for AI skills. - 9.7% of postings mentioned AI in some form, but two-thirds of those mentions were contextual rather than a skill requirement. - Contextual mentions included hiring-process disclosures, company descriptions and job-board features. - Among postings that did ask for AI skills, 80% listed those skills as a requirement rather than a preferred qualification. - Postings asking for AI skills carried a 6.6% higher median salary than other postings in the same occupation. The finding is descriptive, not causal. - 13 postings, or 0.3%, told candidates not to use AI during the hiring process. - Data entry clerks ranked highest on the ILO index among 427 occupations, with a score of 0.70. - None of 200 sampled data entry clerk postings asked for AI skills. - Bookkeepers, payroll clerks, loan officers and office clerks showed similar results, each at 1% or below. - Copywriters had the highest AI-skill demand in the study at 22.5% of postings. - Web developers followed at 11.5%, and software developers at 10.5%. - Copywriters ranked 12th by exposure score, not near the top of the exposure list. - The correlation between ILO exposure score and the share of postings asking for AI was weak, with Spearman rho = 0.26. - Highly exposed occupations were 6.7 times more likely to ask for AI skills than low-exposure office jobs such as HR managers, attorneys and civil engineers, at 4.2% versus 0.6%. - Physical occupations such as registered nurses, truck drivers and electricians had no postings asking for AI skills. - Mango Thrive pulled the data from Coresignal’s multi-source job postings database. - Each posting was scanned with a two-tier dictionary of AI terms and then reviewed in context to separate requirements, preferences, contextual mentions, restrictions on AI use and false positives. - A second reviewer independently blind-checked 100 matching postings. - That review produced 93.2% initial agreement and 98.1% agreement after one classification rule was clarified. - A further 100 non-matching postings were reviewed in full, and no missed AI requirements were found. - The balanced sample means the aggregate figures describe the occupations studied, not the full U.S. labor market. - The full study, methodology, occupation-level tables and complete findings are available in the full study.

Between the lines: - The results point to a mismatch between AI-related job risk and AI-related hiring language. - Employers may be exposed to AI disruption without yet translating that exposure into explicit AI skill requirements. - The strongest AI-skill demand appears in knowledge work tied more directly to digital production than in the roles most exposed on the index.

What's next: - Mango Thrive’s dataset and occupation-level tables give recruiters, job seekers and researchers a baseline for tracking whether AI-skill demand expands over time. - Future job-posting studies can use the same method to see whether AI skill requests spread beyond a few tech-adjacent occupations.

The bottom line: - AI exposure in the workplace is not yet showing up as broad demand for AI skills in job ads.