Surviving the AI Summer

When I was in grad school in the 1990s, we talked about “surviving the AI winter.” Explosive, transformative progress in AI had been promised, but it wasn’t happening. Funders were fed up, and futurologists were getting frustrated and impatient. Of course it would all happen around 2022 — but we didn’t know that in 1995. We needed patience, and strategies for getting through the downturn.

Fast forward to today, and we now have the opposite problem: AI is explosively successful and popular. So much so that in the research world, everything else is getting squeezed out.

I have a friend at an industrial research lab who has often hired my PhD students as summer interns. My students always have a great experience, and the student and mentor get a peer-reviewed paper. I reached out to him a few months ago — I have another great PhD student. Are you interested in an intern? His response was: Do they work on AI? Right now his research organization is only allowed to work on AI.

OK, AI is important and timely. It’s going to take a whole lot of smart people to make this technology work well and fairly. We are just at the beginning, and the hype is only slightly exaggerated. But AI isn’t the only thing. We still need people to research all the other disciplines of computing. But it’s getting harder to do that work.

The “AI Summer” is a burst of interest and money directed to AI so enormous that is blotting out all other activity. Yes, AI is important. But so are ed tech, info vis, social computing, ubicomp — whether or not they use AI. A corporate researcher visiting GT recently told me that we should only train our grad students in AI, because there will be no jobs for them in other disciplines of computing. No corporate jobs, anyway. The sad thing is, they may be mostly right. But what happens if we stop doing work in all the other areas of applied computing? Who will train the next generation of computing researchers in non-AI areas when the hype dies down and some balance returns? What is the cost of not advancing all these other fields of computing for the length of the AI summer?

The AI winter was long. Patience was needed. The AI summer is also going to be long. But those of us who see value in non-AI work need to stick with our values. Balance and sanity will return, if we are patient.

How do non-AI computing researchers survive the AI summer? Leave me a comment.