Abstract: In this talk, I will discuss where the AI & ML ball is today (much closer to the goal than it was even a decade ago) along with some key areas of the research field that the ball will have to cover to reach the goal. In particular, I will focus on the following three research topics: (1) fully exploiting pervasive, heterogeneous, unlinked, messy data; (2) making deep reinforcement learning and control methods work reliably in one shot; and (3) fully personalizing conversational interactive assistants. In the end, there is not just one goal for AI & ML, but many, and these three research areas -- of many possible -- represent key opportunities for a variety of future AI & ML breakthroughs.

A brief introduction to ""Focus on the goal, not the ball": A few years ago, I had the joy of watching my 5-year-old son's soccer game every week, where an extremely cute clump of children would move amoeba-like across the field with the ball hidden somewhere among their 20 feet. Witnessing the lack of collective 5-year-old strategy, I advised my son to stand between the goal and the ball, reasoning that the ball would have to go through his part of the field to reach the goal.

Interestingly, the children's soccer field and the research field of Artificial Intelligence (AI) and Machine Learning (ML) have a lot in common. The AI & ML ball changes its name every decade and the researchers clump around it -- often as much out of enthusiasm as career prudence. But the ball is not the goal and AI & ML in general could benefit from incentivizing a more diversified research strategy. There are always the big, beefy, way-too-tall-for-their-age players who should be chasing the ball. We cannot win without them. But there is also room for the smaller strategic players to position themselves between the ball (what is currently working) and the goal (fully autonomous AI & ML) to facilitate a win for everyone.

Short Bio: Scott Sanner is an Assistant Professor in Industrial Engineering and Cross-appointed in Computer Science at the University of Toronto. He has held previous positions at Oregon State University and National ICT Australia (NICTA) with an adjunct position at the Australian National University. Scott earned a PhD in Computer Science from the University of Toronto (2008), an MS in Computer Science from Stanford University (2002), and a double BS in Computer Science and Electrical and Computer Engineering from Carnegie Mellon University (1999). Scott is currently an Associate Editor for the Artificial Intelligence Journal (AIJ), the Journal of Artificial Intelligence Research (JAIR), and the Machine Learning Journal (MLJ). Scott was a co-recipient of paper awards from the AI Journal (2014), Transport Research Board (2016), and CPAIOR (2018) and a recipient of a Google Faculty Research Award (2020).