AI for Requirements Engineering: Industry adoption and Practitioner perspectives

Human-AI collaboration is the preferred model in practice: The survey found that 54.4% of AI-assisted RE techniques use Human-AI Collaboration (HAIC), while full automation accounts for only 5.4% and passive AI validation just 4.4-6.2%. This indicates practitioners trust AI most when it actively supports human decision-making rather than replacing or passively validating human choices.

AI adoption in RE is already widespread but unevenly distributed: 58.2% of the 55 surveyed practitioners reported already using AI in Requirements Engineering, yet the paper notes "limited research has examined AI adoption in RE," suggesting a gap between industry practice and academic understanding of how AI is being deployed in this critical software engineering phase.

Practitioners are optimistic but governance frameworks are lacking: While 69.1% of respondents view AI's impact on RE as positive or very positive, the authors emphasize the need for "RE-specific HAIC frameworks along with robust and responsible AI governance as AI adoption in RE grows," indicating that current safeguards may not keep pace with adoption rates.

AI works best as an active partner, not a passive checker: Practitioners significantly prefer AI that actively collaborates with them over AI that simply validates their work, suggesting that responsible AI implementation should focus on transparent, interactive support rather than black-box oversight or autonomous decision-making.