Tweet thread of work from my students/collaborators at @AIIDEconference, what better way to spend your Friday afternoon? #aiide20

Nathan Sturtevant ( @nathansttt) and I explored the effects that even single level changes could have on player experience in "The Unexpected Consequence of Incremental Design Changes". This is part of an ongoing project on Exhaustive PCG for co-design.
https://webdocs.cs.ualberta.ca/~nathanst/papers/sturtevant2020incremental.pdf

A huge collection of graduate students worked with me this Spring on the "Image-to-Level" system, which automatically converts images to game levels. The idea is to introduce a low-effort way for non-experts to employ machine learning for PCG.

http://guzdial.com/2020/aiide20-poster.pdf

If you liked my "Morai Maker" work on human-AI collaboration for level design, then you'll also likely be into our new "Mechanic Maker" demo for human-AI collaboration for game mechanics/rule design. (Project lead by an undergrad, Vardan Saini!)

http://guzdial.com/2020/aiide20-demo1.pdf

Faraz Khadivpour ( @KhadivpourFaraz) and I developed a way to automatically identify which training instance was most "responsible" for the learned behaviour of a Deep RL agent, to be presented at @exag20xx #exag20

https://arxiv.org/pdf/2010.01676.pdf

Keeping the #exag20 train going, Nazanin Yousefzadeh Khameneh ( @Nazaninyz) has work on learning an "entity embedding" via a VAE (think word embedding but for game entities/sprites). We think this can be a huge aide for DNN mechanic reasoning work.

https://arxiv.org/pdf/2010.01685.pdf

Last of #exag20, Adrián Gonzalez ( @EtreSerBe) has work on automatically altering Pokemon images to different types with a VAE.

This is just the beginning of his PhD work, looking at making a general visual PCGML system. He's also part of the AIIDE DC!
https://arxiv.org/pdf/2010.01681.pdf

Finally, another collaboration with @nathansttt from our co-supervised PhD student Kristen Yu (also in the AIIDE DC!), investigating the various definitions of "Quest" in prior work, to be presented at the Interactive Narrative Technology Workshop #int20

http://guzdial.com/2020/yu_INT2020.pdf

And that's it! Thanks so much for reading. Lots of follow-ups on these projects and entirely new ones happening now. Happy to answer questions on any this.