Upcoming research on Stack Overflow and across the Stack Exchange network (May 2025)

Question

Today, I shared an update on the blog about the research we will be conducting over the next quarter. TL;DR – this quarter’s research focus is as follows:

• Exploring new content types around collaborative knowledge building and expert consultation and feedback.
• Exploring users’ mental model and how that affects the way someone posts to Stack Overflow. How do discussions and questions differ? Should we simplify our entry point? Is there a place for every question?
• Evaluating search usability and how we might help users find answers faster. But also how might users find relevant content to consume and engage with more easily.
• Usability tests for features like chat, coding challenges, etc.

I want to personally thank you all for all of the input and engagement with research within Meta and chat over the last few months. I see us agreeing that changes need to be made across the network and I am excited that we are entering this journey together. I am excited to see what we learn in the next quarter and how we, together, can give Stack Overflow the future it deserves.

My usual reminder—we need highly engaged research participants. If you are someone who is comfortable answering questions about your experience please sign up for research invitations! While our previous quarter’s research was heavily targeted towards Stack Overflow users, our need to talk with Stack Exchange users is coming! We hope to speak with you soon.

• V2Blast: Related previous post: Upcoming research on Stack Overflow and across the Stack Exchange network (Feb 2025)
• Piper: @TimurShtatland These tasks are performed by two different job functions.
• V2Blast: @TimurShtatland: The researchers at Stack don't fix bugs. They... do research. That's their job. Different staff can do different things at the same time.

Answer 1

A lesson from the blog post:

Help seekers struggle to get their problem solved on Stack Overflow. Writing a good question takes too much time in today’s world especially when most don’t know how to write in the Stack Overflow quality standard.

Could this be an opportunity for AI? Could AI make it easier to ask a good question on Stack Exchange sites? Could AI give seekers feedback on their question, suggest ways they could improve it, and ask them questions that readers would likely have and help them incorporate it into the question?

• Piper: Yes! We have you question assistant on Stack Overflow, but it trained on SO data and would be hard to scale to SE networks. It has improved question asking success rates which is always a good thing!
• user400654: Unfortunately, the current iteration more often than not just gives poor advice.
• EmmaBee: @KevinB do you have specific examples of the poor advice given? Maybe this is posted on meta already, but curious if you have some examples you can point to. (do you have personal experience with the advice, or have you seen the advice others were given?)
• user400654: I haven't been back to it in a while, but this was my first experience with it: meta.stackoverflow.com/questions/432630/… The gist of my issue with the question assistant is it does all this work behind the scenes trying to figure out what guidance might be useful, which is great, but then it just dumps it into a summarization prompt and generates a mess with useful information intertwined within it.