Sign in

Matthew Guzdial

@matthewguz.bsky.social
985 followers 353 following 938 posts

Associate professor @ University of Alberta and Canada CIFAR AI Chair @ Alberta Machine Intelligence Institute. Games, machine learning, and creativity | he/him

PostsRepliesMedia
Reposted by Matthew Guzdial
Slashie @slashie.bsky.social · 3h
I've been given the honor of giving the closing talk at this year's #Roguelike Celebration @roguelike.club, joining a stellar speaker lineup. I'd share my talk's title, but it would go over bsky's character limit! Check out the full program: www.roguelike.club/event2026.html
The Decades-Long Design Experiment That Escapes Permadeath and Finds a New Form Every Run
I’ve been making roguelikes for more than two decades, and I’ve come to realize that I haven’t really been making dozens of separate games: instead I’ve been running one very long design experiment. This talk follows the ideas that have survived that experiment: mechanics, systems, interfaces, world-generation techniques, and questions that disappear for years only to emerge again and sometimes find a working shape.

This isn’t a story of steady progress toward better games, rather an exploration of how curiosity, iteration, failure, and unfinished work accumulate over time. (true) Roguelikes are built around the idea that knowledge persists across runs even when characters don’t, I think game development can work the same way.
1268
Matthew Guzdial @matthewguz.bsky.social · 6h
Thanks for sharing! I think the general thrust seems nice but can you say why you don’t teach the MDP formalism? I find it makes things way faster and I can usually cover it in like 15 mins with examples.
120
Matthew Guzdial @matthewguz.bsky.social · 8h
They just did a big hiring push for it, including some big names, expect another round of marketing push for it soon.
020
Matthew Guzdial @matthewguz.bsky.social · 10h
And currently with Genie!
120
Reposted by Matthew Guzdial
Sebastian Deterding @codingconduct.cc · 18h
Come work with us! The Dyson School of Design Engineering at @imperialcollegeldn.bsky.social is hiring for an Assistant Professor in Fair and Responsible AI. www.nature.com/naturecareer...
nature.com
Assistant Professor in Fair and Responsible Artificial Intelligence - South Kensington Campus job with Imperial College London (ICL) | 12862563
We are seeking candidates who develop novel AI methods and systems that are technically rigorous, trustworthy and deployable in practice
133
Reposted by Matthew Guzdial
Interactive Arts Alberta @iaalberta.bsky.social · 05/10/2026
Interested in learning more about cooperatives or worker centric business models but not with a studio? Check out the upcoming Prairie Winter Studio Jam!
022
Matthew Guzdial @matthewguz.bsky.social · 05/10/2026
Ah yes, the ACM house style
030
Reposted by Matthew Guzdial
Mike Cook @mtrc.bsky.social · 05/10/2026
Every AI/HCI paper I review now just sounds like this.
Research [25] shows that people [26] enjoy the feeling of winning games [3,5,19]. Enjoying things can lead to longer life expectancy [2], higher economic productivity [9,22,33-40,302] and improve mental health [9]. Yet many games still include losing players [67,4]. A recent study showed that less than half of the players in a Chess [1] match win on average [42]. We have developed an LLM-driven co-creative [66], user-led [11], mixed-initiative [80] assistant for Chess [2], through which players can redefine the rules of Chess [3] using generative AI prompting such that they are the winner of Chess [4]. A study (n=3) showed that 100\% of people felt better, healthier and more self-actualised after using our tool, showing that this may be the way forward for games and human culture more broadly in the future.
30801194
Reposted by Matthew Guzdial
Marlos C. Machado @marloscmachado.bsky.social · 05/10/2026
I'm excited to share Erik Lintunen's work! Mastering Atari 2600 Games with Discovered Options We have a general-purpose option-discovery method that works w/ high-dimensional observations, tabula rasa, using a single stream of experience! It outperforms DreamerV3, Rainbow, and IQN! 1/9
2185
Matthew Guzdial @matthewguz.bsky.social · 05/10/2026
My hypothesis is these are the same thing. Sycophancy as a vector for belief change. Humans have been doing it to get the king’s favour for millennia.
110
Reposted by Matthew Guzdial
Nicholas Grossman @nicholasgrossman.bsky.social · 04/10/2026
AI is now the biggest investment in all of US history as a percentage of GDP. Datacenter construction is up, all other construction is down. "AI Is Squeezing Out the Rest of the Stock Market," reported WSJ 2 days ago. We basically bet the whole economy on the most optimistic visions panning out.
821992473
Reposted by Matthew Guzdial
Lisa Diedrich @lisadiedrich.bsky.social · 03/10/2026
"This is not models going rogue. This is gross incompetence & negligence by a particular company, which actually they could be held criminally liable for." @timnitgebru.blacksky.app cuts to the AI chase.
3493160
Reposted by Matthew Guzdial
MHDante @mhdante.com · 02/10/2026
Hey, there's this game I helped make is 50% off!. Check it out! It's a fun exploration on what it's like to train a neural network. (This AI game has no generative AI in it)
store.steampowered.com
Save 50% on Little Learning Machines on Steam
Welcome to Little Learning Machines—a simulation where you train your real AI companions to solve challenges together, across whimsical islands. Shape their abilities and personalities while leveling ...
041
Matthew Guzdial @matthewguz.bsky.social · 02/10/2026
Fantastic work Si! Love the motion!
110
Reposted by Matthew Guzdial
Mascot Texorcism 🐯 @texsaber.bsky.social · 02/10/2026
I've had this stupid joke in my head for the past 3 weeks, I had to put it into drawing
Theatrical poster for "Twin Peaks: Fire Walk with Me," except the portrait of Laura Palmer is replaced with a Walkling from the game Big Walk, and the film name is replaced with "Big Walk with Me"
3252411633
Reposted by Matthew Guzdial
Timnit Gebru @timnitgebru.blacksky.app · 01/10/2026
I'm not sure if y'all saw a CNN report on how the US had almost started WW3 with China based on fabricated information generated by a chatbot, claiming that a Chinese vessel was carrying components of nuclear weapons. That was an "existential threat" to humanity www.theguardian.com/commentisfre...
theguardian.com
Forget ‘superintelligence’: error-prone AI nearly sparked world war three this month | Timnit Gebru and Emily M Bender
The risks of AI aren’t what we think they are, as a recent security incident between China and the United States reveals
10411169
Reposted by Matthew Guzdial
brendan mordheim @brendanmcleod.dev · 01/10/2026
This month on PAPER LABYRINTHS, we chat with narrative designer and writer Bruno Dias @brunodias.dev about video game writing as a distinct discipline, information theory, weird fiction, and the AmWriting industrial complex. sites.libsyn.com/568595/site/...
01810
Reposted by Matthew Guzdial
Mark J. Nelson @mm-jj-nn.bsky.social · 01/10/2026
We have an opening for a tenured Professor / Associate Professor of Interactive Media "with an emphasis on interactive storytelling, worldbuilding, and/or playful design in the public interest". I'm on the hiring committee and happy to answer questions.
american.wd1.myworkdayjobs.com
Associate or Full Professor of Interactive Media Film and Media Arts School of Communication
American University is a student-centered research institution located in Washington, DC, with highly-ranked schools and colleges, internationally-renowned faculty, and a reputation for creating meani...
12015
Matthew Guzdial @matthewguz.bsky.social · 01/10/2026
I’d happily take the bet on the side of the rate of new games on major digital storefronts not being meaningfully different in 18 months. I kind of think with changes in funding it might actually go down.
110
Reposted by Matthew Guzdial
Marco @mcognetta.bsky.social · 30/09/2026
🚨 [Token][ization] Paper Alert 🚨 Tokenization is a wildly understudied area of language modeling despite it having effects across all of NLP. Over the past ~8 months, 32 (!) tokenizer researchers put together the most comprehensive survey of the field. Check it out!
112935
Reposted by Matthew Guzdial
Amanda Silberling @amanda.omg.lol · 30/09/2026
sometimes writing about bluesky discourse means you're just too deep in the echo chamber but other times bluesky discourse uncovers the unsettling reality that there are minecraft jokes coded into an official government website techcrunch.com/2026/09/29/a...
techcrunch.com
America.gov gets really weird when you ask it about Minecraft, but it's not a glitch | TechCrunch
For the sake of national security, it's a relief to learn that America.gov is not hallucinating to the point that it's penning lengthy poetry.
2175
Reposted by Matthew Guzdial
WIRED @wired.com · 29/09/2026
However, some see the doom talk as a distraction from discussions of the technology’s actual harms. Timnit Gebru, one of the fiercest critics of AI, told WIRED believes the doom talk is about founders making money, not saving humanity. www.wired.com/story/the-bi...
"The funder, the founders, the investors who stand to benefit and profit the most from these companies IPO-ing are saying this, that they have been the ones seeding this narrative going back multiple decades." - Timnit Gebru, AI researcher and prominent critic
37118
Reposted by Matthew Guzdial
Mike Cook @mtrc.bsky.social · 29/09/2026
The schedule's out - I'm speaking at the Roguelike Celebration again this year! My talk is titled "The City, The Street, The Lens" and it's about street photography, collisions in time and space and how to think about procedural generation differently. More here: www.roguelike.club/event2026.html
A photo of a truck by the side of the road in Tokyo. The driver's leg is sticking out of the window resting on the side of the trailer. The title of the talk is written on the truck. It will happen on October 18th, online. Tickets are roguelike.club
17622
Reposted by Matthew Guzdial
Alfred Reinold Baudisch @alfredbaudisch.com · 24/09/2026
Pet project: 1) Advanced PS1/N64 game extraction (+all animations). 2) Website where you can browse and compare models, textures, animations, FMVs, for preservation. Already have FF7, Mega Man Legends 1 and 2 and MediEvil fully ripped. #FF7 #FinalFantasy #FFVII #PS1 #PSX #RetroGame #N64
20878299
Matthew Guzdial @matthewguz.bsky.social · 28/09/2026
We've got info up on the recommended hotels for AIIDE 2026, including the special rate for the AIIDE block at Mercure Belo Horizonte Lourdes!
010
Matthew Guzdial @matthewguz.bsky.social · 27/09/2026
Jackbox Party Pack 12’s MegaPals is the ESP Game with scoring based on single-linkage clustering, a little gift for me and anyone else who needs lecture examples.
020
Matthew Guzdial @matthewguz.bsky.social · 26/09/2026
They’re so, so desperate for large scale real world data from consumer’s perspectives. They will do anything short of just paying people to get it.
0112
Reposted by Matthew Guzdial
Cat Manning @catacalypto.bsky.social · 26/09/2026
was literally just referencing this—the “who is having the most fun, the player, the designer, or the computer” question—earlier today. it’s such a core concern when making something heavily systemic or relies a lot on procgen
317827
Matthew Guzdial @matthewguz.bsky.social · 26/09/2026
The manatees don’t even want to eat the bacon. Takes a lot of arguing to convince them it’s a kind of kelp. But worth the trouble!!
020
Reposted by Matthew Guzdial
brendan mordheim @brendanmcleod.dev · 25/09/2026
Star Wars Zero Company has been nominated for two Golden Joystick awards - PC Game of the Year, and Best Supporting Performer (Vic Michaelis!). Vote now! www.gamesradar.com/goldenjoysti...
gamesradar.com
Golden Joystick Awards 2026: Vote NOW for your games of the year!
Did your favorite games make the shortlist? Find out below
0182
Reposted by Matthew Guzdial
Mark Riedl @markriedl.bsky.social · 25/09/2026
Someone asked Muse to dump its entire virtual machine file system, and it did
35417
Matthew Guzdial @matthewguz.bsky.social · 23/09/2026
Ooh good one! I might steal that.
010
Matthew Guzdial @matthewguz.bsky.social · 23/09/2026
Digging myself out of a month deep e-mail debt, thank goodness it will never get that bad again! (behind Matthew an even larger e-mail debt looms menacingly)
180
Matthew Guzdial @matthewguz.bsky.social · 22/09/2026
(the whole blog post flying over my head) wow I can’t believe you’re coming to Canada and not saying hi. (lovely post Mike! sorry things have been so rough lately)
110
Reposted by Matthew Guzdial
Mike Cook @mtrc.bsky.social · 22/09/2026
New Post: Why I Love AI Next month it'll be twenty years since I went to university, learned to program, and studied AI. I wrote about why I loved it, why I still love it, and why, unfortunately, it no longer exists. www.possibilityspace.org/blog/posts/i...
32554164
Matthew Guzdial @matthewguz.bsky.social · 22/09/2026
I know, just a joke!
100
Matthew Guzdial @matthewguz.bsky.social · 22/09/2026
Wait but I thought these folks were true believers of the possibility of recursive self improvement?
150
Reposted by Matthew Guzdial
John Gallagher @johnrgallagher.bsky.social · 20/09/2026
Free preview of AI Through the Experts’ Eyes! (introduction) upittpress.org/wp-content/u...
upittpress.org
173
Matthew Guzdial @matthewguz.bsky.social · 20/09/2026
Thanks a bunch! I uploaded the recipe to a google doc here docs.google.com/document/d/1...
docs.google.com
Low Fodmap Carrot Cake Recipe
Ingredients 2 cups (260g) chopped walnuts (1 cup for cake, 1 cup for garnish)* 1 and 1/2 cups (300g) maple syrup 1/2 cup (100g) granulated sugar 1 cup (226g/240ml) vegetable oil (or melted coconut o...
110
Matthew Guzdial @matthewguz.bsky.social · 19/09/2026
I estimate within the year we’ll see a major collapse, I think this is just the early rumblings before the avalanche.
100
Matthew Guzdial @matthewguz.bsky.social · 19/09/2026
If you’re Sheev and the only way you know how to feel joy is bullying force wielders and you’ve just drastically reduced your number of possible victims the inquisitors start sounding like a really good idea.
020
Reposted by Matthew Guzdial
Shannon Vallor @shannonvallor.bsky.social · 19/09/2026
This is going around as essential reading and for very good reason. Honestly the most incisive and full summary of the longtermism/x-risk pathology I have seen. Every stanza is a gem, and the closing is genius, but my personal favorite, because I have had this exact conversation almost verbatim:
[nitter.app post excerpt from banteg]

>someone asks where the one-in-a-million number came from
> expert judgement
>which expert
>us
>how calibrated
> extremely thoughtfully

›reduce estimate to one in ten million to be conservative
›still beats curing cancer by 38 orders of magnitude
›epistemic robustness achieved
87925
Matthew Guzdial @matthewguz.bsky.social · 18/09/2026
Oh no, get well soon!
011
Reposted by Matthew Guzdial
AIIDE @aiide.bsky.social · 18/09/2026
We invite paper authors to submit companion software, dataset, or tool to the optional Artifact Evaluation. ​Accepted artifacts earn special recognition at the conference! Learn more: sites.google.com/view/aiide20...
011
Matthew Guzdial @matthewguz.bsky.social · 18/09/2026
He uses all the resources to cast the greatest thaumaturgical work ever devised to fully remove Britain from existence, regretfully ending the series.
010
Reposted by Matthew Guzdial
Gautam Kamath @gautamkamath.com · 16/09/2026
Nihar Shah did a heroic experiment for TMLR: he spent 20-25 hours over two weeks interviewing authors of seemingly low-quality submissions about their own papers. He confirmed what we all suspected: people submitting these papers have *no idea* what is going on in them.
5293117
Reposted by Matthew Guzdial
brendan mordheim @brendanmcleod.dev · 13/09/2026
with Band of Blades back in the TTRPG collective imaginary, may I recommend OATHSWORN, a complete Band of Blades actual play set in an original world? Eldritch skies, body horror, political infighting, comedic interludes - we've got it all. oathswornpodcast.com
oathswornpodcast.com
OATHSWORN
A Band of Blades Actual Play Podcast
04122
Matthew Guzdial @matthewguz.bsky.social · 13/09/2026
Congratulations Cat! Best of luck with whatever comes next, can’t wait to see what it is!
010
Reposted by Matthew Guzdial
Jake Gold @jacob.gold · 12/09/2026
There's no slowing down frontier models by regulation that wouldn't lead to regulatory capture. But we could simply force all AI model companies to release their model weights openly. It's the world's data after all and nothing would slow their funding/progress more effectively and universally.
1125442
Reposted by Matthew Guzdial
Katie Mack @astrokatie.com · 12/09/2026
“these solutions are announced in a rush, leaving no time for a proper writeup, the isolation of new methods & ideas, & citing relevant previous work of others. … We are witnessing a general threat to intellectual work, with misalignment between the outcome of the use of AI and its initial purpose.”
9758177