Welcome to our next 'Mini Minds' installment! As I mentioned before, we're still figuring out what we want this format to be, so you can expect a bit of tinkering over the coming months. Our first mini was a short audio blogpost of sorts, and today's mini is a mini interview. I chatted with Matt Crosby. He's a postdoc at Imperial College London and has been spearheading a super cool project called the Animal AI Olympics. (If you recall, this is something that Marta Halina and I touched on briefly in our last...
Welcome to our next 'Mini Minds' installment!
As I mentioned before, we're still figuring out what we want this format to be, so you can expect a bit of tinkering over the coming months. Our first mini was a short audio blogpost of sorts, and today's mini is a mini interview. I chatted with Matt Crosby. He's a postdoc at Imperial College London and has been spearheading a super cool project called the Animal AI Olympics. (If you recall, this is something that Marta Halina and I touched on briefly in our last episode, but it seemed intriguing enough to merit a longer look.) The basic idea behind the project, as Matt and I discuss, was to have a bunch of artificial agents—submitted by teams from around the world—compete in a gauntlet of tests. The wrinkle was that these weren't the kinds of tests usually given to artificial systems. They were tests usually given to animals: some involved avoiding obstacles; others involved remembering locations; still others involved solving physical puzzles, like the celebrated "trap tube" task. This all sounds like good fun—and no doubt it was. But there were also deeper motivations for the project. Matt and I talk about those. We also talk a bit about how the contestants performed, about whether he was impressed, and about his team's plans for the next iteration of the Olympics, to be held in 2021.
Enjoy the mini—and thanks for listening!
A transcript of this interview is available here.
Notes and links
2:23 – An article by Dr. Crosby and colleagues about the Animal AI Olympics and the motivations behind it.
3:45 – An article about recent successes in getting AIs to play Atari games.
5:45 – For details about the Animal AI environment and the types of tasks use, see here.
8:28 – A little more about object permanence.
12:48 – Variants of the "trap tube" task have been widely used in animal cognition research. Here's an example of a paper using it in apes, and here is a video of a dog attempting this kind of task.
14:15 – A blogpost announcing the results of the competition.
For more info about the Animal AI Olympics, check out: http://www.animalaiolympics.com/
For more about Matt Crosby and his work, see:
https://www.mdcrosby.com/
@MaCroPhilosophy
Many Minds is a project of the Diverse Intelligences Summer Institute (DISI) (https://www.diverseintelligencessummer.com/), which is made possible by a generous grant from the Templeton World Charity Foundation to UCLA. It is hosted by Kensy Cooperrider, with creative support from DISI Directors Erica Cartmill and Jacob Foster, and Associate Director Hilda Loury. Our artwork is by Ben Oldroyd (https://www.mayhilldesigns.co.uk/). Our transcripts are created by Sarah Dopierala (https://sarahdopierala.wordpress.com/).
You can subscribe to Many Minds on Apple, Stitcher, Spotify, Pocket Casts, Google Play—or wherever you like to listen to podcasts.
We welcome your comments, questions, and suggestions. Feel free to email us at: manymindspodcast@gmail.com.
For updates about the show, follow us on Twitter: @ManyMindsPod.
