Job ads show AI is reshaping skill demand: postings that require AI increasingly list human-centered skills such as analytical thinking, teamwork and resilience, and firms with more AI roles advertise fewer substitutable skills; some complementary skills (notably resilience) also command modest posted-salary premia.
Complement or Substitute? How AI Increases the Demand for Human Skills
Using ~30 million online job ads from 2018–2024 across four countries, the paper finds that AI-related roles—and workplaces with higher AI prevalence—are associated with greater demand for human-centered complementary skills (analytical thinking, resilience, teamwork) and, in some cases, modest wage premiums.
<title>Abstract</title> Artificial Intelligence (AI) is transforming the nature of work, yet there is limited empirical evidence on how it affects demand for human skills. This paper examines whether AI adoption increases the prevalence and value of human capabilities that complement technical AI skills, such as analytical thinking, resilience, or ethical judgment, within and beyond AI-intensive job roles. Using a dataset of nearly 30 million job postings from the US, the UK and Australia, between 2018 and 2024, we distinguish between internal effects (within AI roles) and external effects (in non-AI roles) across companies, industries, and regions. This paper has three main findings. First, we find that AI-intensive roles are significantly more likely to require complementary non-technical capabilities, such as analytical thinking, resilience, and digital literacy. Second, these complementary skills are associated with meaningful wage premiums, particularly in managerial, sales or finance roles working with AI. Third, we show that AI diffusion has potential spillover effects: as AI adoption rises within companies, industries, and regions, demand for complementary skills increases even in non-AI roles while demand for substitutable skills – summarisation, translation or customer service – decreases. These trends hold across geographies, including the United States, United Kingdom, and Australia, confirming the robustness of our findings. Together, these findings indicate that AI is not simply replacing tasks or requiring more AI developer skills, it may be transforming workforce skill requirements to favor human attributes that enhance collaboration with intelligent systems.
Summary
Main Finding
AI adoption increases demand for non-technical, complementary human capabilities (e.g., analytical thinking, resilience, digital literacy, ethical judgment) both inside AI-intensive roles and, via spillovers, in non-AI roles across firms, industries and regions. These complementary skills are associated with meaningful advertised wage premiums (especially in managerial, sales, and finance roles), while demand for more directly substitutable skills (e.g., summarisation, translation, routine customer-service tasks) falls as AI diffuses. Results are robust across the United States, United Kingdom and Australia (2018–2024).
Key Points
- Three headline results:
- AI-intensive roles are significantly more likely to list complementary non-technical capabilities (analytical thinking, resilience, digital literacy, ethical judgment).
- These complementary skills correlate with measurable wage premiums, particularly in managerial, sales, and finance occupations that work with AI.
- AI diffusion produces spillovers: as adoption rises within companies/industries/regions, demand for complementary skills increases even in non-AI roles, while demand for substitutable, routine task skills decreases.
- Complementary skills cited: analytical reasoning, problem-solving, resilience/adaptability, ethical judgment, digital literacy, interpersonal/coordination capabilities.
- Substitutable skills cited: summarisation, translation, routine customer service tasks (skills more easily automated by current AI).
- Geographic robustness: findings replicate across three advanced-economy labor markets (US, UK, Australia) and across the 2018–2024 period.
- The evidence implies AI is reshaping job-content and skill demand beyond simply expanding demand for AI developers or replacing tasks.
Data & Methods
- Data: nearly 30 million online job postings from the US, UK and Australia spanning 2018–2024.
- Skill measurement: text-based extraction/parsing of job ads to identify skill mentions and classify skills as complementary (human-capability-oriented) or substitutable (AI-automatable tasks).
- Role classification: identification of AI-intensive roles (roles that explicitly require AI/ML skills or work closely with AI systems) versus non-AI roles.
- Empirical strategy (as reported): comparative analysis of skill prevalence and advertised wages across:
- within-role contrasts (AI-intensive vs non-AI roles),
- within-firm/industry/region changes over time to capture diffusion/spillovers.
- Outcome measures: prevalence of skill mentions in postings; advertised wage/salary information where available to estimate associated wage premiums.
- Robustness: cross-country comparisons and checks that trends hold across different geographies and occupational groups.
- Interpretation caveat: results document strong associations and patterns consistent with complementarity and spillovers; causal attribution to AI requires further identification strategies (e.g., quasi-experimental variation) beyond these descriptive/panel analyses.
Implications for AI Economics
- Labor demand: AI adoption reshapes demand away from routinizable, automatable tasks toward human capabilities that complement AI—policy and firm responses should emphasize those skills.
- Wage structure: premium for complementary interpersonal/analytical capabilities may increase returns to workers who can partner with AI, potentially altering within-occupation pay spreads (not just between-occupation changes).
- Human capital policy: education, training and reskilling programs should focus on analytical reasoning, adaptability, ethical judgment, and digital literacy rather than only on technical AI development skills.
- Employer strategy: firms may prioritize hiring/training for collaboration skills (management, coordination, judgment) as they deploy AI across functions, and redesign jobs to leverage human–AI complementarities.
- Regional and industry effects: spillovers mean AI diffusion in one firm/industry can raise complementary-skills demand locally, suggesting place-based workforce strategies and regional labor-market monitoring.
- Research directions: need for causal identification of AI’s effect on wages and long-term career outcomes, evaluation of effective training interventions, and analysis of distributional consequences across demographic groups and skill levels.
Assessment
Paper Typecorrelational
Evidence Strengthmedium — The study leverages an exceptionally large, multi-country dataset (~30 million job ads) and reports consistent patterns across the US, UK, Australia and New Zealand, strengthening external validity of the observed associations; however, the analysis is observational with no plausibly exogenous variation to establish causality, faces measurement issues (proprietary skill tagging, binary AI-role definition), and is vulnerable to confounding (firm heterogeneity, selection into posting, reverse causality), limiting causal inference.
Methods Rigormedium — The authors use appropriate statistical models for binary and fractional outcomes (logistic, fractional-logit) and regressions on log salaries, control for many covariates, and conduct analyses at multiple aggregation levels and across countries; but reliance on proprietary skill taxonomies, a simple 'any-AI-skill' rule, limited wage analyses for subgroups, absence of panel causal designs (e.g., difference-in-differences, event studies, IVs), and limited treatment of potential endogeneity reduce methodological rigor.
SamplePrimary analysis uses ~10 million US online job postings (2018–2024) from Lightcast/Burning Glass (scraped from >65,000 sources) with job title, occupation, posted salary, and a Lightcast-generated skill list; extended sample includes ~30 million postings covering the US, UK, Australia and New Zealand (2018–2024). AI roles are defined by presence of any AI/ML skill; complementary and substitutable skills are coded into clusters based on prior literature. Analyses include cross-sectional company-, industry-, and region-level aggregates.
Themeshuman_ai_collab skills_training
IdentificationDescriptive and multivariate association analysis using skill-based classification of job postings: AI roles are identified if a posting lists any Lightcast 'Artificial Intelligence and Machine Learning' skill; complementary and substitutable skill clusters are coded from skill lists. The authors estimate logistic/fractional-logistic regressions (and linear regressions on log posted salaries) with controls (experience, education, industry, occupation, etc.) and present cross-sectional company-, industry-, and region-level associations; no quasi-experimental variation, instrumental variables, or longitudinal causal identification strategy is used.
GeneralizabilityJob-posting data measure employer demand, not realized employment or worker tasks; advertised skills may differ from on-the-job requirements., Data come from high-income, English-speaking countries (US, UK, Australia, New Zealand) — findings may not generalize to low-/middle-income countries or non-English labour markets., Lightcast/Burning Glass proprietary skill extraction and the binary ‘any-AI-skill’ rule may misclassify roles and skills (measurement error)., Posted salaries are incomplete or noisy and may not reflect final wages or compensating differentials; wage analyses are limited in subgroup power., Cross-sectional associations are vulnerable to firm heterogeneity and reverse causality (firms that hire complementary-skilled workers may be more likely to adopt AI), limiting causal generalizability across contexts and time., Rapid evolution in AI (post-2024 developments) may change the patterns observed during 2018–2024.
Claims (6)
| Claim | Direction | Confidence | Outcome | Details |
|---|---|---|---|---|
| AI-intensive roles are significantly more likely to require complementary non-technical capabilities such as analytical thinking, resilience, and digital literacy. Skill Acquisition | positive | high | prevalence (requirement) of complementary non-technical capabilities in AI-intensive job postings |
n=30000000
0.3
|
| Complementary (non-technical) skills are associated with meaningful wage premiums, particularly in managerial, sales, or finance roles working with AI. Wages | positive | medium | wage premium associated with complementary skills (salary level differences) |
n=30000000
0.18
|
| As AI adoption rises within companies, industries, and regions, demand for complementary skills increases even in non-AI roles. Skill Acquisition | positive | medium | demand for complementary skills in non-AI roles (frequency of skill requirements) |
n=30000000
0.18
|
| As AI adoption rises, demand for substitutable skills—such as summarisation, translation, or customer service—decreases. Skill Obsolescence | negative | medium | demand for substitutable skills (frequency of skill mentions) |
n=30000000
0.18
|
| These trends (increased demand for complementary skills and decreased demand for substitutable skills) hold across geographies including the United States, United Kingdom, and Australia, demonstrating robustness. Skill Acquisition | positive | high | consistency of skill-demand trends across countries |
n=30000000
0.3
|
| AI is not simply replacing tasks or only requiring more AI developer skills; it may be transforming workforce skill requirements to favor human attributes that enhance collaboration with intelligent systems. Skill Acquisition | mixed | medium | shift in workforce skill requirements toward human attributes that complement AI |
n=30000000
0.18
|