One of the (very valid) complaints about how I determined the best AI school was that I only looked at publications from AAAI/IJCAI. Each of the major subfields of AI have their own particular conferences that people will often prefer to send work to over AAAI. ML, probably the most dynamic and exciting part of today’s AI landscape, is better served by NIPS and ICML. Robotics, by ICRA. I’ve heard the case made that agent stuff is better at AAMAS.

So, what’s the point of AAAI? Arguments about fostering interdisciplinary work go out the window when there are a dozen parallel tracks. What we call AI is so vast and tenuous that interdisciplinary work might be a fool’s errand anyway. Furthermore, why attend a track that just has “pretty good” work in robotics?

The only thing I can think of in AAAI’s favor is that it’s important (for job searches, &c.) that there’s a recognizable AI conference that everyone knows is “good”. But if that ML paper was so important, why didn’t you publish it at NIPS?