Strateegia + GPT beats AI-only for requirements elicitation in new study
Hybrid human-AI approach scores highest on quality criteria from ISO standard.
A new empirical study by Manoel Salgado Neto, Alan Araujo, and Ronnie de Souza Santos (arXiv:2606.24060) tackles a key question in requirements engineering: how does AI support affect collaborative requirements elicitation? The researchers ran a mixed-method controlled experiment comparing four approaches: (1) traditional collaborative elicitation without AI, (2) collaborative elicitation using the Strateegia platform with its GPT-powered Writer applet, (3) direct requirements generation using a large language model, and (4) generating requirements from collaborative discussion transcripts using an LLM. All artifacts were evaluated using quality criteria derived from the ISO/IEC/IEEE 29148 standard.
The results clearly favored hybrid approaches. Combining stakeholder collaboration with AI-supported synthesis produced the highest-rated requirements artifacts and was perceived as clearer and easier to execute than traditional collaborative elicitation. The study provides empirical evidence that generative AI can effectively transform collaboratively generated knowledge into structured requirements documentation while preserving the value of stakeholder participation. The authors discuss implications for human-AI collaboration in requirements engineering, suggesting that AI is most powerful when augmenting—not replacing—human interaction.
- Hybrid approach (Strateegia + GPT) scored highest on ISO/IEC/IEEE 29148 quality criteria.
- Participants found AI-supported collaborative elicitation clearer and easier than traditional methods.
- Direct LLM generation and transcript-only LLM approaches underperformed compared to human-AI collaboration.
Why It Matters
Shows that AI works best as a synthesis partner, not a replacement, for complex knowledge work like requirements engineering.