Artificial Intelligence papers are taking over many software conferences. We have recently seen this extreme case:
— JB (@JBezivin) February 2, 2024
Even if the above example itself is a kind of “clickbait” (papers are from this Mining Challenge track were focused on DevGPT, a curated dataset of developer-ChatGPT conversations), I still think it’s a realistic scenario of what could happen in the future if many conferences follow the current trend of accepting more and more AI-related papers at the expenses of other topics that were the core of the conference some years ago.
You could argue that this is part of the normal evolution of research. There have always been occasional research bubbles in software engineering that generate new topics that dominate for a while. And often, they disappear shortly afterward. Does anybody remember the “Aspect bubble” when we saw hundreds of papers applying aspect-orientation to every single software engineering task? (for more fun on the paradoxical success of aspect-orientation, you should read this).
I don’t think AI will disappear anytime soon. And of course, I think AI is a giant contribution to software engineering. But I also fear we’re being too carried away. And investing too much researcher effort on this topic at the expenses of other core important topics in our communities.
So I wonder, Shouldn’t we make sure that conferences keep some kind of “topic diversity” by putting in place some measures for the conservation of old-fashioned topics?
We know the importance of biodiversity in our ecosystems. I think keeping some type of “topicdiversity” is equally relevant. Or we may lose some improvements that could the basis of a new wave of progress in the future.
If you agree, we should then discuss how to actually implement conservation measures. An option would be to have a track for core topics (called them “old-fashion” papers if you want 😆 ).
And the discussion should also go beyond publications and span to funding calls. If we want research on these topics to continue, publishing the results is the last step. Funding them is the first one!