Matthew Guzdial
@matthewguz.bsky.social
Associate professor @ University of Alberta and Canada CIFAR AI Chair @ Alberta Machine Intelligence Institute.
Games, machine learning, and creativity | he/him
Games, machine learning, and creativity | he/him
Reposted by Matthew Guzdial
Nonviolence is, in fact, working. This administration is weaker than it was a year ago. More and more of the public is becoming galvanized against it; its agents are being impeded.
Do you honestly think people in the civil rights movement and other nonviolent movements faced no violence themselves? They practiced nonviolence because it works, because the oppressor wants a shooting war, which is the war it will win
How's your non-violent working Mr Gandhi stancil? They will kill each and everyone of you right if they had the chance and you will still be writing Kumbaya.
January 24, 2026 at 9:53 PM
Nonviolence is, in fact, working. This administration is weaker than it was a year ago. More and more of the public is becoming galvanized against it; its agents are being impeded.
Reposted by Matthew Guzdial
It's the Polaris 2025 reports! They've come! They're here! It's a bumper crop of five, count'em FIVE papers! We got comedy, we got death, we got cringe, we got trust... something for the whole team.
Piping hot game design knowledge and insight await:
polarisgamedesign.com/reports/
Piping hot game design knowledge and insight await:
polarisgamedesign.com/reports/
January 21, 2026 at 9:42 PM
It's the Polaris 2025 reports! They've come! They're here! It's a bumper crop of five, count'em FIVE papers! We got comedy, we got death, we got cringe, we got trust... something for the whole team.
Piping hot game design knowledge and insight await:
polarisgamedesign.com/reports/
Piping hot game design knowledge and insight await:
polarisgamedesign.com/reports/
Reposted by Matthew Guzdial
NEW: I spoke to Jonathan Blow about his ambitious new puzzle game and why it took ten years to make. Then, days after the announcement, three designers came out denouncing Blow for his views on Trump (pro), vaccines (anti), and more. Here's the whole messy story: www.bloomberg.com/news/newslet...
A Controversial Video-Game Designer Returns With a New Game and More Controversy
The announcement of ‘Order of the Sinking Star’ was met with criticism over the political views of its creator Jonathan Blow
www.bloomberg.com
January 22, 2026 at 9:45 PM
NEW: I spoke to Jonathan Blow about his ambitious new puzzle game and why it took ten years to make. Then, days after the announcement, three designers came out denouncing Blow for his views on Trump (pro), vaccines (anti), and more. Here's the whole messy story: www.bloomberg.com/news/newslet...
Reposted by Matthew Guzdial
This year's Polaris game design papers are up and our group worked on Mechanical Comedy in Games, as in games where the comedy comes from players performing the jokes. Really pleased with what we learned (and glad I thought to pitch the topic).
polarisgamedesign.com/2025/mechani...
polarisgamedesign.com/2025/mechani...
Mechanical Comedy In Games 🤡 😂
A framework for designing mechanical comedy
polarisgamedesign.com
January 22, 2026 at 7:16 PM
This year's Polaris game design papers are up and our group worked on Mechanical Comedy in Games, as in games where the comedy comes from players performing the jokes. Really pleased with what we learned (and glad I thought to pitch the topic).
polarisgamedesign.com/2025/mechani...
polarisgamedesign.com/2025/mechani...
Reposted by Matthew Guzdial
Update: Now 120 cities are holding Minneapolis General Strike solidarity actions
paydayreport.com/minneap/
paydayreport.com/minneap/
MAP: Minneapolis General Strike Movement Spreads to 120 Cities on Friday
At least 80 cities plan to hold actions in solidarity with the Minneapolis General Strike.
paydayreport.com
January 22, 2026 at 12:28 PM
Update: Now 120 cities are holding Minneapolis General Strike solidarity actions
paydayreport.com/minneap/
paydayreport.com/minneap/
Reposted by Matthew Guzdial
Are low-quality survey respondents bots?
Five years of mixed methods research reveals problematic respondents are often humans trying to get paid for surveys that they do not quality for (based on language proficiency, location, experience, etc.):
doi.org/10.1177/1745...
Five years of mixed methods research reveals problematic respondents are often humans trying to get paid for surveys that they do not quality for (based on language proficiency, location, experience, etc.):
doi.org/10.1177/1745...
!["In the classic version, most people pull the lever to save five people at the expense of one innocent bystander. However, in the footbridge version, most switch their preference, choosing not to push one man to his death to save five others."
"These findings have been replicated across hundreds of thousands of participants and numerous cultures ..., but not among the problematic respondents in our studies. Instead, most problematic respondents sacrifice one man to save five in both versions of the dilemma...."
"...problematic participants overwhelmingly tend to respond 'yes,' regardless of context. This pattern of acquiescence... likely reflects incentives in the online-research ecosystem. On many participant platforms (especially market-research platforms), people qualify for studies by endorsing various behaviors. [So acquiescence may be] less indicative of bots and more of people motivated to increase their survey-taking profits."](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:jf4udtyuylqocyyrqdwqoxuf/bafkreigi5x4gzow5xulknhek2oz7evc3i6qygihwk4ggwph4a7kg43ca2y@jpeg)
!["From a large pool of low-quality respondents..., we randomly invited... respondents [to] interviews [which] revealed that all [interviewees] were, indeed, human. Most participants appeared seated at long rows of Formica desks, with snippets of Telugu and Hindi speech in the background. Using Zoom’s chat function (because of participants’ limited English), we asked questions that could indicate positivity bias. Specifically, we examined whether they would say 'yes' to questions to which 'yes' could not possibly be an accurate answer, revealing their tendency to indiscriminately respond 'yes' regardless of question content."
"...we conducted ...more of these interviews [with people taking a survey] while sharing their screens. ...participants were often in rooms with multiple laptops and people working together. [They often] changed places on camera in the middle of a survey, [used] screen-sharing software, ...VPNs, ...Google translate, [and exhibited] patterns of acquiescence".](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:jf4udtyuylqocyyrqdwqoxuf/bafkreiaeef6jggg6tyr55jbmpo5ostkcce5nusygiigpxuu4gjw2ng3pam@jpeg)
January 22, 2026 at 12:28 PM
Are low-quality survey respondents bots?
Five years of mixed methods research reveals problematic respondents are often humans trying to get paid for surveys that they do not quality for (based on language proficiency, location, experience, etc.):
doi.org/10.1177/1745...
Five years of mixed methods research reveals problematic respondents are often humans trying to get paid for surveys that they do not quality for (based on language proficiency, location, experience, etc.):
doi.org/10.1177/1745...
Reposted by Matthew Guzdial
The disconnect between management and employee on AI usage is big. At the same time, CEOs are reporting they aren't seeing any revenue growth from AI.
🤷
www.wsj.com/lifestyle/wo...
🤷
www.wsj.com/lifestyle/wo...
CEOs Say AI Is Making Work More Efficient. Employees Tell a Different Story.
How much time workers say the technology saves them on the job is vastly different from what executives report.
www.wsj.com
January 21, 2026 at 2:07 PM
The disconnect between management and employee on AI usage is big. At the same time, CEOs are reporting they aren't seeing any revenue growth from AI.
🤷
www.wsj.com/lifestyle/wo...
🤷
www.wsj.com/lifestyle/wo...
Reposted by Matthew Guzdial
Hey folks, I am looking for more illustration and character design work this year. I have over a decade of experience freelancing for tabletop, card and video games.
You can check out more work and client testimonials on my website: www.sifsweetman.com
Contact: [email protected]
You can check out more work and client testimonials on my website: www.sifsweetman.com
Contact: [email protected]
January 6, 2026 at 6:08 PM
Hey folks, I am looking for more illustration and character design work this year. I have over a decade of experience freelancing for tabletop, card and video games.
You can check out more work and client testimonials on my website: www.sifsweetman.com
Contact: [email protected]
You can check out more work and client testimonials on my website: www.sifsweetman.com
Contact: [email protected]
Reposted by Matthew Guzdial
Patronize as many of these businesses as you can before and after Friday.
When we started this list of Minnesota businesses taking part in the Jan. 23 economic blackout we didn't realize quite what an undertaking it would be, but we are now at more than 250 and counting.
Keep 'em coming!
Keep 'em coming!
List of Minnesota businesses closing for Jan. 23 anti-ICE strike
First announced by faith leaders and unions on Jan. 13, Minnesota businesses have begun announcing plans to join the action.
bringmethenews.com
January 21, 2026 at 1:28 AM
Patronize as many of these businesses as you can before and after Friday.
A senior professor years ago told me they were doing “metaverse” research, and I cautioned against it as it was obviously doomed. But they were bullish, saying how could Zuckerberg spend all this money if there was nothing to it.
The division has lost over $77 billion since its inception in 2020.
Meta Lays Off Thousands of VR Workers as Zuckerberg's Vision Fails
The company has laid off some 1,500 virtual reality developers as part of a broader shift toward AI wearable devices.
trib.al
January 18, 2026 at 5:05 PM
A senior professor years ago told me they were doing “metaverse” research, and I cautioned against it as it was obviously doomed. But they were bullish, saying how could Zuckerberg spend all this money if there was nothing to it.
Reposted by Matthew Guzdial
Would love to speak to any regular international GDC attendees who are skipping this year's show (or were going to attend and have now cancelled) due to the current political climate. DMs here or [email protected], thank you!
January 17, 2026 at 12:43 AM
Would love to speak to any regular international GDC attendees who are skipping this year's show (or were going to attend and have now cancelled) due to the current political climate. DMs here or [email protected], thank you!
Reposted by Matthew Guzdial
Shared my thoughts on Canada's AI strategy for this new piece in @thewalrus.ca
The tendency to frame AI as inevitable & autonomous "obscures what we’re really talking about, which is people—in business, in government, in civil society—and the decisions they make."
thewalrus.ca/evan-solomon...
The tendency to frame AI as inevitable & autonomous "obscures what we’re really talking about, which is people—in business, in government, in civil society—and the decisions they make."
thewalrus.ca/evan-solomon...
Evan Solomon Wants Canada to Trust AI. Can We Trust Evan Solomon? | The Walrus
The journalist-turned-minister says the tech will make us richer and regulation should be “light”
thewalrus.ca
January 16, 2026 at 4:13 PM
Shared my thoughts on Canada's AI strategy for this new piece in @thewalrus.ca
The tendency to frame AI as inevitable & autonomous "obscures what we’re really talking about, which is people—in business, in government, in civil society—and the decisions they make."
thewalrus.ca/evan-solomon...
The tendency to frame AI as inevitable & autonomous "obscures what we’re really talking about, which is people—in business, in government, in civil society—and the decisions they make."
thewalrus.ca/evan-solomon...
It’s fun to spend the morning with a journalist correcting all the popularly-believed lies about AI. It’s wild to be reminded that some people don’t know we’re in an AI bubble/summer or that it has happened before.
January 15, 2026 at 4:57 PM
It’s fun to spend the morning with a journalist correcting all the popularly-believed lies about AI. It’s wild to be reminded that some people don’t know we’re in an AI bubble/summer or that it has happened before.
Reposted by Matthew Guzdial
Something else that really struck me in this report is this paragraph. "AI is doing things for students that they used to enjoy."
January 14, 2026 at 5:53 PM
Something else that really struck me in this report is this paragraph. "AI is doing things for students that they used to enjoy."
Reposted by Matthew Guzdial
The only answer to crunch is building worker power through unions. Executives would crunch even if they 100% believed it was bad for business because executives are motivated by wanton cruelty and a desire to exercise power as much as anything.
January 14, 2026 at 12:05 PM
The only answer to crunch is building worker power through unions. Executives would crunch even if they 100% believed it was bad for business because executives are motivated by wanton cruelty and a desire to exercise power as much as anything.
Reposted by Matthew Guzdial
Literally gave a talk yesterday and this was my final slide. There are two paths.
One is to teach LLMs to do proper reasoning & planning. Can't rule this out.
Another is to leverage known, powerful reasoning and planning algorithms and make them work in language space. Can't rule this out either.
One is to teach LLMs to do proper reasoning & planning. Can't rule this out.
Another is to leverage known, powerful reasoning and planning algorithms and make them work in language space. Can't rule this out either.
January 13, 2026 at 4:15 PM
Literally gave a talk yesterday and this was my final slide. There are two paths.
One is to teach LLMs to do proper reasoning & planning. Can't rule this out.
Another is to leverage known, powerful reasoning and planning algorithms and make them work in language space. Can't rule this out either.
One is to teach LLMs to do proper reasoning & planning. Can't rule this out.
Another is to leverage known, powerful reasoning and planning algorithms and make them work in language space. Can't rule this out either.
Reposted by Matthew Guzdial
We are thrilled to announce that AIIDE-26 will be held at the Federal University of Minas Gerais, Belo Horizonte, MG, Brazil from November 9-13, 2026! Paper abstracts due June 19 with full submissions due June 26, 2026. sites.google.com/view/aiide20...
AIIDE 2026
The 22nd AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE 2026) is the next in an annual series of conferences showcasing interdisciplinary research on modeling,...
sites.google.com
January 12, 2026 at 5:18 PM
We are thrilled to announce that AIIDE-26 will be held at the Federal University of Minas Gerais, Belo Horizonte, MG, Brazil from November 9-13, 2026! Paper abstracts due June 19 with full submissions due June 26, 2026. sites.google.com/view/aiide20...
Reposted by Matthew Guzdial
🚨[Pls repost!] Agentic AI is stress-testing Explainable AI. We need to fix it. That's why I'm thrilled to announce the 6th Human-Centered Explainable AI (HCXAI) workshop at CHI 2026 in Barcelona! 🚀
📝 2-5 page single column papers (excluding refs)
🗓️ Deadline: Feb 19, 2026
🔗 hcxai.jimdosite.com
1/n
📝 2-5 page single column papers (excluding refs)
🗓️ Deadline: Feb 19, 2026
🔗 hcxai.jimdosite.com
1/n
January 12, 2026 at 8:58 PM
🚨[Pls repost!] Agentic AI is stress-testing Explainable AI. We need to fix it. That's why I'm thrilled to announce the 6th Human-Centered Explainable AI (HCXAI) workshop at CHI 2026 in Barcelona! 🚀
📝 2-5 page single column papers (excluding refs)
🗓️ Deadline: Feb 19, 2026
🔗 hcxai.jimdosite.com
1/n
📝 2-5 page single column papers (excluding refs)
🗓️ Deadline: Feb 19, 2026
🔗 hcxai.jimdosite.com
1/n
Reposted by Matthew Guzdial
I’m super excited and honored to help organize the very first edition of AIIDE outside North America! @matthewguz.bsky.social and I are planning a memorable event at UFMG in Belo Horizonte, Brazil. We're going with the appropriate theme of "New Grounds". Please submit your best work and join us!
We are thrilled to announce that AIIDE-26 will be held at the Federal University of Minas Gerais, Belo Horizonte, MG, Brazil from November 9-13, 2026! Paper abstracts due June 19 with full submissions due June 26, 2026. sites.google.com/view/aiide20...
AIIDE 2026
The 22nd AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE 2026) is the next in an annual series of conferences showcasing interdisciplinary research on modeling,...
sites.google.com
January 12, 2026 at 6:04 PM
I’m super excited and honored to help organize the very first edition of AIIDE outside North America! @matthewguz.bsky.social and I are planning a memorable event at UFMG in Belo Horizonte, Brazil. We're going with the appropriate theme of "New Grounds". Please submit your best work and join us!
So thrilled to be taking AIIDE outside of North America for the first time! Joined by @lucasnfe.bsky.social as Program Chair we're going with the very appropriate theme of "New Grounds". Please submit all your best work and join us this November!
We are thrilled to announce that AIIDE-26 will be held at the Federal University of Minas Gerais, Belo Horizonte, MG, Brazil from November 9-13, 2026! Paper abstracts due June 19 with full submissions due June 26, 2026. sites.google.com/view/aiide20...
AIIDE 2026
The 22nd AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE 2026) is the next in an annual series of conferences showcasing interdisciplinary research on modeling,...
sites.google.com
January 12, 2026 at 5:22 PM
So thrilled to be taking AIIDE outside of North America for the first time! Joined by @lucasnfe.bsky.social as Program Chair we're going with the very appropriate theme of "New Grounds". Please submit all your best work and join us this November!
Reposted by Matthew Guzdial
yeah; it really comes down to gently saying
1. don't hurt others
2. don't hurt yourself
3. don't hurt the organization
4. don't hurt me
Get through those successfully and you've got someone setup for success.
1. don't hurt others
2. don't hurt yourself
3. don't hurt the organization
4. don't hurt me
Get through those successfully and you've got someone setup for success.
January 12, 2026 at 1:38 AM
yeah; it really comes down to gently saying
1. don't hurt others
2. don't hurt yourself
3. don't hurt the organization
4. don't hurt me
Get through those successfully and you've got someone setup for success.
1. don't hurt others
2. don't hurt yourself
3. don't hurt the organization
4. don't hurt me
Get through those successfully and you've got someone setup for success.
In any leadership position I have found that the number one time commitment is just stopping people from hurting themselves. I call this my "Lemmings" theory of management.
January 12, 2026 at 1:23 AM
In any leadership position I have found that the number one time commitment is just stopping people from hurting themselves. I call this my "Lemmings" theory of management.
Reposted by Matthew Guzdial
@bottleofpennies.bsky.social
and I dived into this year's AIIDE
@emilyhalina.bsky.social and @ravyn.ca
on our podcast!
open.spotify.com/episode/2ugT...
and I dived into this year's AIIDE
@emilyhalina.bsky.social and @ravyn.ca
on our podcast!
open.spotify.com/episode/2ugT...
Episode 25: We went to a conference again
open.spotify.com
January 11, 2026 at 1:57 AM
@bottleofpennies.bsky.social
and I dived into this year's AIIDE
@emilyhalina.bsky.social and @ravyn.ca
on our podcast!
open.spotify.com/episode/2ugT...
and I dived into this year's AIIDE
@emilyhalina.bsky.social and @ravyn.ca
on our podcast!
open.spotify.com/episode/2ugT...
Reposted by Matthew Guzdial
I strongly suggest this read on the topic and LLM antropomorphism:
The LLMentalist Effect: how chat-based Large Language Models replicate the mechanisms of a psychic’s con
softwarecrisis.dev/letters/llme...
The LLMentalist Effect: how chat-based Large Language Models replicate the mechanisms of a psychic’s con
softwarecrisis.dev/letters/llme...
January 8, 2026 at 2:51 PM
I strongly suggest this read on the topic and LLM antropomorphism:
The LLMentalist Effect: how chat-based Large Language Models replicate the mechanisms of a psychic’s con
softwarecrisis.dev/letters/llme...
The LLMentalist Effect: how chat-based Large Language Models replicate the mechanisms of a psychic’s con
softwarecrisis.dev/letters/llme...
Reposted by Matthew Guzdial
It's well past time for Canadian government officials to get off of X, @jameswsthomson.com writes. This latest — disgusting — incident is just one more reason.
Canadian government officials, what are you still doing on X in 2026?
The platform's own chatbot started posting sexualized material of children and other people who couldn't consent. With no moral bottom in sight, it's well past time our officials found other ways to c...
www.nationalobserver.com
January 2, 2026 at 8:45 PM
It's well past time for Canadian government officials to get off of X, @jameswsthomson.com writes. This latest — disgusting — incident is just one more reason.