Impact of AI on the Software Industry

Determine the effects of artificial intelligence (AI), including large language models such as ChatGPT, Mistral, and Perplexity, on the software industry to ascertain how AI will change software development practices, structures, and outcomes.

Background

The paper extends economic complexity analysis to the digital sector using open-source software data from GitHub, constructing a Software Economic Complexity Index and examining relatedness dynamics. In the Discussion, the authors consider the emerging role of LLMs in software development and note that their broader effects on the software industry remain uncertain.

This uncertainty is situated in a context where LLMs appear to provide better programming advice in languages with more training data, potentially influencing ubiquitous languages and countries with lower software complexity. The authors flag that understanding AI's impact is essential for interpreting future changes in software capabilities and economic complexity.

References

Though it is unclear how AI will affect the software industry(Quispe and Grijalba, 2024), we can already observe that LLMs are better at providing programming advice in languages for which they have more training data (del Rio-Chanona et al. 2023), which are likely to be ubiquitous languages.

— The Software Complexity of Nations  (2407.13880 - Juhász et al., 2024) in Discussion

These results therefore rule out era shifts in the stable share larger than about $\pm$0.23, but they are too wide to certify that smaller shifts are absent. Put plainly, the stable factor's weight grew after ChatGPT's release by about as much as it would have grown anyway, and any true era effect, if one exists, is too small for this design to detect.

— Chance, Persistent Advantage, and the Generative-AI Era in Open-Source Package Careers  (2609.09687 - Ibrahim et al., 9 Sep 2026) in Section 5, Results, subsection “RQ3: No detectable break in the generative-AI era”

Our study does not establish how practitioners respond when these choices are constrained.

— The Psychological Costs of Artificial Intelligence Adoption in Software Engineering  (2609.03456 - Alami et al., 3 Sep 2026) in Section 7.1, “Boundary Conditions of the Findings”