Penetration of artificial intelligence into the software development life cycle: An empirical labor market analysis
O. V. Stoyanova, Ivan S. Okuskov · Business Informatics · 30 Jun 2026
Read original sourceOpen access
Where this sits in the debate
Direction of the finding as read from the abstract by an AI model (AI-written).
Key takeaway AI-written
Analyzing 182,447 Russian IT job postings (hh.ru, March-December 2025) across 19 software-development roles, finds AI-skill penetration into early-phase software work (requirements analysis, planning) is lower than in later phases like design and development, with weighted AI-competency penetration ranging from 0.34% (support) to 1.70% (design). It's a granular labor-market data point on where, within a single knowledge-work pipeline, AI skills are actually showing up in job postings versus where they aren't yet.
Abstract
The integration of artificial intelligence (AI) into information technologies is reshaping competency requirements for IT specialists. However, AI competency penetration into the early phases of the Software Development Life Cycle (SDLC) – requirements analysis and planning – remains underexplored. Existing studies either rely on static competency models lacking empirical validation or analyze the labor market at an aggregated occupational level without differentiation across SDLC functional clusters. The research objective is to empirically test the hypothesis of AI competency penetration into the skill sets associated with early SDLC phases. The scientific novelty lies in the methodological transition from static competency models to a process-oriented analysis. Roles are cross-referenced with SDLC phases through weighted clustering with normative alignment to SWEBOK v3.0, BABOK v3.0, and ISO/IEC/IEEE 12207:2017. The empirical basis comprises 182,447 unique IT job postings from the hh.ru platform for the period March–December 2025, covering 19 SDLC-relevant roles. A reference list of 307 AI competencies was compiled via LLM-based preliminary screening followed by expert verification. Eight functional clusters are identified. Weighted demand S(s) is distributed as follows: analytics – 16.0%, architecture – 16.0%, development – 13.6%, management – 12.2%, documentation – 12.2%, testing – 12.0%, support – 9.1%, design – 8.9%. Weighted penetration P(s) ranges from support (0.34%) to design (1.70%). The hypothesis is confirmed: penetration in the Analytics cluster (0.69%) corresponds to the market average, while in the Management cluster (0.76%) it exceeds it, though early SDLC phases lag behind the Design, Development, and Architecture clusters in absolute terms. The findings are applicable to corporate early-warning strategies, educational curriculum revision, and updating professional standards in the context of the proliferation of generative AI.
About this publication
Academic journal. Published in an academic journal — typically peer-reviewed before publication, though indexing alone doesn't guarantee review depth.
- Published
- 2026 (4 months ago)
- Citations
- 0 (OpenAlex)
0 citations recorded by OpenAlex — too few 2026 papers in this collection for a fair comparison.
Related by theme, not necessarily by method or conclusion.
- The Effects of Generative AI on High Skilled Work: Evidence from Three Field Experiments with Software Developers
SSRN · 2024 - GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models
arXiv · 2023 - Experimental evidence on the productivity effects of generative artificial intelligence
Science · 2023 - The simple macroeconomics of AI
Economic Policy · 2024
Overview assembled automatically from collected metadata. How this site is built.