I recently started a series of surveys on LinkedIn. We just launched the third survey (and it would be great if you could take 30 seconds to participate!). Our goal is to take some of the hype out of AI in requirements engineering. (Also don’t forget to sign up for the panel on AI and requirements on September 19, that I will participate in.)
We already extracted a few interesting insights. For instance the fact that AI currently is not that great in the role of a co-interviewer. Contrast this with the insight that AI is already doing a decent job at transcription and summarizing. Why?
Why AI Transcription Works and AI Co-Interviewers Don’t
For one, the infrastructure for transcription had already been established. Since Covid, online meetings are widely accepted. Integrating a transcription service into a meeting tool is easy, as the AI just has to listen. Contrast that with AI Co-Interviewers, who would have to engage interactively. This is much more challenging, as it could become very annoying if it does not work perfectly.
Checking the quality of requirements is both the currently most available and also most valuable use case.
Another insight: Checking requirements quality is perceives as one of the most valuable use cases — more valuable than writing requirements, for instance. But this might be due to the fact that AI does this already really well. In fact, AI-based requirements quality checking has been around ten years ago, and rule-based quality checking for much longer.
Updated Survey Results at SE-Trends
We maintain the survey results online at SE-Trends (in English) and will also send another message, once the survey series is complete.
I hope that this information is useful for you. And if you like this content, then you can say thank you by participating in the surveys yourself and by joining the discussion on LinkedIn!
