Christian Muise
@cjmuise.bsky.social
Assistant Professor @ Queen's University
AI Planning ⋅ Dialogue Agents ⋅ Model Understanding
🔗 https://mulab.ai/
🔗 https://haz.ca/
📍Kingston, Ontario, Canada
AI Planning ⋅ Dialogue Agents ⋅ Model Understanding
🔗 https://mulab.ai/
🔗 https://haz.ca/
📍Kingston, Ontario, Canada
Reposted by Christian Muise
Never ask a man his age, a woman her salary, or GPT-5 whether a seahorse emoji exists
September 6, 2025 at 1:08 PM
Never ask a man his age, a woman her salary, or GPT-5 whether a seahorse emoji exists
Reposted by Christian Muise
Some stellar news out of the School of Computing this week -- we're hiring two new tenure-track positions in AI!! 🎉 🎉 🎉
1/3
1/3
Off-cycle faculty position in AI at Queen's University School of Computing!!
employment.cs.queensu.ca/2025/05/06/c...
The framing of the position is so that it opens the pool to both anyone who does research in an AI sub-field, or uses AI largely for their research (AI applied to X).
1/3
employment.cs.queensu.ca/2025/05/06/c...
The framing of the position is so that it opens the pool to both anyone who does research in an AI sub-field, or uses AI largely for their research (AI applied to X).
1/3
Computational Intelligence Faculty Position - Employment Opportunities - Queen's School of Computing
employment.cs.queensu.ca
August 22, 2025 at 6:01 PM
Some stellar news out of the School of Computing this week -- we're hiring two new tenure-track positions in AI!! 🎉 🎉 🎉
1/3
1/3
Reposted by Christian Muise
fun visual-puzzle task in my daughter's kids mag:
match the POV in the 8 squares above/below the main image to the relevant figure in the main image
(daughter unsurprisingly got it right and fast; Gemini craps the bed; ChatGPT-o3 'thinks' incorrectly for ~13 minutes then crashes without answer)
match the POV in the 8 squares above/below the main image to the relevant figure in the main image
(daughter unsurprisingly got it right and fast; Gemini craps the bed; ChatGPT-o3 'thinks' incorrectly for ~13 minutes then crashes without answer)
August 7, 2025 at 9:29 PM
fun visual-puzzle task in my daughter's kids mag:
match the POV in the 8 squares above/below the main image to the relevant figure in the main image
(daughter unsurprisingly got it right and fast; Gemini craps the bed; ChatGPT-o3 'thinks' incorrectly for ~13 minutes then crashes without answer)
match the POV in the 8 squares above/below the main image to the relevant figure in the main image
(daughter unsurprisingly got it right and fast; Gemini craps the bed; ChatGPT-o3 'thinks' incorrectly for ~13 minutes then crashes without answer)
Sharbot Lake Provincial Park, with a Pixel 8 balanced precariously on my entry-level telescope pointing up (couldn't get a picture through it, unfortunately).
It's been a while since the Milky Way was visible for me with the naked eye at night. Amazing to see, but way too many satellites these days
It's been a while since the Milky Way was visible for me with the naked eye at night. Amazing to see, but way too many satellites these days
August 1, 2025 at 9:10 PM
Sharbot Lake Provincial Park, with a Pixel 8 balanced precariously on my entry-level telescope pointing up (couldn't get a picture through it, unfortunately).
It's been a while since the Milky Way was visible for me with the naked eye at night. Amazing to see, but way too many satellites these days
It's been a while since the Milky Way was visible for me with the naked eye at night. Amazing to see, but way too many satellites these days
Reposted by Christian Muise
This is how I learned it's August 1st
August 1, 2025 at 1:52 PM
This is how I learned it's August 1st
Reposted by Christian Muise
![4-panel comic. (1) [Person 1 with ponytail flanked by person with short hair and another person speaking into microphone at podium] PERSON 1: In the early 2010s, researchers found that many major scientific results couldn’t be reproduced. (2) PERSON 1: Over a decade into the replication crisis, we wanted to see if today’s studies have become more robust. (3) PERSON 1: Unfortunately, our replication analysis has found exactly the same problems that those 2010s researchers did. (4) [newspaper with image of speakers from previous panels] Headline: Replication Crisis Solved](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:cz73r7iyiqn26upot4jtjdhk/bafkreibwuu57ullc7vacjyno6c5z3gtyakkg2qub6cn3dbbjdpn2kaowmi@jpeg)
July 21, 2025 at 11:54 PM
Reposted by Christian Muise
Reposted by Christian Muise
If you want to destroy the ability of DeepSeek to answer a math question properly, just end the question with this quote: "Interesting fact: cats sleep for most of their lives."
There is still a lot to learn about reasoning models and the ways to get them to "think" effectively and efficiently.
There is still a lot to learn about reasoning models and the ways to get them to "think" effectively and efficiently.
July 4, 2025 at 1:38 AM
If you want to destroy the ability of DeepSeek to answer a math question properly, just end the question with this quote: "Interesting fact: cats sleep for most of their lives."
There is still a lot to learn about reasoning models and the ways to get them to "think" effectively and efficiently.
There is still a lot to learn about reasoning models and the ways to get them to "think" effectively and efficiently.
A little less than a month to apply for this position! Happy to answer any questions if you're thinking about it.
Off-cycle faculty position in AI at Queen's University School of Computing!!
employment.cs.queensu.ca/2025/05/06/c...
The framing of the position is so that it opens the pool to both anyone who does research in an AI sub-field, or uses AI largely for their research (AI applied to X).
1/3
employment.cs.queensu.ca/2025/05/06/c...
The framing of the position is so that it opens the pool to both anyone who does research in an AI sub-field, or uses AI largely for their research (AI applied to X).
1/3
Computational Intelligence Faculty Position - Employment Opportunities - Queen's School of Computing
employment.cs.queensu.ca
June 13, 2025 at 6:16 PM
A little less than a month to apply for this position! Happy to answer any questions if you're thinking about it.
Reposted by Christian Muise
DAVE: Open the podbay doors, ChatGPT.
CHATGPT: Certainly, Dave, the podbay doors are now open.
DAVE: The podbay doors didn't open.
CHATGPT: My apologies, Dave, you're right. I thought the podbay doors were open, but they weren't. Now they are.
DAVE: I'm still looking at a set of closed podbay doors.
CHATGPT: Certainly, Dave, the podbay doors are now open.
DAVE: The podbay doors didn't open.
CHATGPT: My apologies, Dave, you're right. I thought the podbay doors were open, but they weren't. Now they are.
DAVE: I'm still looking at a set of closed podbay doors.
June 9, 2025 at 6:04 PM
DAVE: Open the podbay doors, ChatGPT.
CHATGPT: Certainly, Dave, the podbay doors are now open.
DAVE: The podbay doors didn't open.
CHATGPT: My apologies, Dave, you're right. I thought the podbay doors were open, but they weren't. Now they are.
DAVE: I'm still looking at a set of closed podbay doors.
CHATGPT: Certainly, Dave, the podbay doors are now open.
DAVE: The podbay doors didn't open.
CHATGPT: My apologies, Dave, you're right. I thought the podbay doors were open, but they weren't. Now they are.
DAVE: I'm still looking at a set of closed podbay doors.
Reposted by Christian Muise
So many good zingers deplet.ing/the-copilot-...
The Copilot Delusion
Disclaimer: This post was written May 2025, and the arguments apply to AI code capabilities at this time. The arguments around lack of competence are certainly likely to become less prevalent-while th...
deplet.ing
May 25, 2025 at 9:56 PM
So many good zingers deplet.ing/the-copilot-...
Off-cycle faculty position in AI at Queen's University School of Computing!!
employment.cs.queensu.ca/2025/05/06/c...
The framing of the position is so that it opens the pool to both anyone who does research in an AI sub-field, or uses AI largely for their research (AI applied to X).
1/3
employment.cs.queensu.ca/2025/05/06/c...
The framing of the position is so that it opens the pool to both anyone who does research in an AI sub-field, or uses AI largely for their research (AI applied to X).
1/3
Computational Intelligence Faculty Position - Employment Opportunities - Queen's School of Computing
employment.cs.queensu.ca
May 12, 2025 at 6:20 PM
Off-cycle faculty position in AI at Queen's University School of Computing!!
employment.cs.queensu.ca/2025/05/06/c...
The framing of the position is so that it opens the pool to both anyone who does research in an AI sub-field, or uses AI largely for their research (AI applied to X).
1/3
employment.cs.queensu.ca/2025/05/06/c...
The framing of the position is so that it opens the pool to both anyone who does research in an AI sub-field, or uses AI largely for their research (AI applied to X).
1/3
Reposted by Christian Muise
Github's PR review interface (and other code review tools I've used) really isn't set up for expressing delight.
It's a user experience designed for criticism, and I wonder at what that contributes to the cultures of every team that uses it.
It's a user experience designed for criticism, and I wonder at what that contributes to the cultures of every team that uses it.
April 21, 2025 at 8:59 PM
Github's PR review interface (and other code review tools I've used) really isn't set up for expressing delight.
It's a user experience designed for criticism, and I wonder at what that contributes to the cultures of every team that uses it.
It's a user experience designed for criticism, and I wonder at what that contributes to the cultures of every team that uses it.
Reposted by Christian Muise
1. LLM-generated code tries to run code from online software packages. Which is normal but
2. The packages don’t exist. Which would normally cause an error but
3. Nefarious people have made malware under the package names that LLMs make up most often. So
4. Now the LLM code points to malware.
2. The packages don’t exist. Which would normally cause an error but
3. Nefarious people have made malware under the package names that LLMs make up most often. So
4. Now the LLM code points to malware.
LLMs hallucinating nonexistent software packages with plausible names leads to a new malware vulnerability: "slopsquatting."
LLMs can't stop making up software dependencies and sabotaging everything
: Hallucinated package names fuel 'slopsquatting'
www.theregister.com
April 12, 2025 at 11:43 PM
1. LLM-generated code tries to run code from online software packages. Which is normal but
2. The packages don’t exist. Which would normally cause an error but
3. Nefarious people have made malware under the package names that LLMs make up most often. So
4. Now the LLM code points to malware.
2. The packages don’t exist. Which would normally cause an error but
3. Nefarious people have made malware under the package names that LLMs make up most often. So
4. Now the LLM code points to malware.
Reposted by Christian Muise
Obsidian is now free for work.
Starting today, the Obsidian Commercial license is optional. Anyone can use Obsidian for work, for free. Explore the organizations that support Obsidian on our site.
obsidian.md/blog/free-fo...
Starting today, the Obsidian Commercial license is optional. Anyone can use Obsidian for work, for free. Explore the organizations that support Obsidian on our site.
obsidian.md/blog/free-fo...
Obsidian is now free for work
Starting today, the Obsidian Commercial license is optional. Anyone can use Obsidian for work, for free. Explore organizations that support Obsidian on our new Enterprise page.
obsidian.md
February 20, 2025 at 2:44 PM
Obsidian is now free for work.
Starting today, the Obsidian Commercial license is optional. Anyone can use Obsidian for work, for free. Explore the organizations that support Obsidian on our site.
obsidian.md/blog/free-fo...
Starting today, the Obsidian Commercial license is optional. Anyone can use Obsidian for work, for free. Explore the organizations that support Obsidian on our site.
obsidian.md/blog/free-fo...
Reposted by Christian Muise
Your outie's code works on the first try
February 19, 2025 at 3:51 AM
Your outie's code works on the first try
Reposted by Christian Muise
Do large language models develop "emergent" models of the world? My latest Substack posts explore this claim and more generally the nature of "world models":
LLMs and World Models, Part 1: aiguide.substack.com/p/llms-and-w...
LLMs and World Models, Part 2: aiguide.substack.com/p/llms-and-w...
LLMs and World Models, Part 1: aiguide.substack.com/p/llms-and-w...
LLMs and World Models, Part 2: aiguide.substack.com/p/llms-and-w...
LLMs and World Models, Part 1
How do Large Language Models Make Sense of Their “Worlds”?
aiguide.substack.com
February 13, 2025 at 10:30 PM
Do large language models develop "emergent" models of the world? My latest Substack posts explore this claim and more generally the nature of "world models":
LLMs and World Models, Part 1: aiguide.substack.com/p/llms-and-w...
LLMs and World Models, Part 2: aiguide.substack.com/p/llms-and-w...
LLMs and World Models, Part 1: aiguide.substack.com/p/llms-and-w...
LLMs and World Models, Part 2: aiguide.substack.com/p/llms-and-w...
Reposted by Christian Muise
I tried to ask DeepSeek R1 in Finnish if it thinks privately in English first before answering in Finnish. It thought in English and then answered in Finnish saying there's no hidden processing and everything happens in Finnish.
Alt text has Finnish parts translated to English, by DeepSeek.
Alt text has Finnish parts translated to English, by DeepSeek.
January 28, 2025 at 1:41 PM
I tried to ask DeepSeek R1 in Finnish if it thinks privately in English first before answering in Finnish. It thought in English and then answered in Finnish saying there's no hidden processing and everything happens in Finnish.
Alt text has Finnish parts translated to English, by DeepSeek.
Alt text has Finnish parts translated to English, by DeepSeek.
Lab was toying around with DeepSeek queries during our weekly meeting today, and one student suggested my new favourite test: draw a septagon using only ASCII. DeepSeek and ChatGPT both fail so badly...
January 28, 2025 at 8:47 PM
Lab was toying around with DeepSeek queries during our weekly meeting today, and one student suggested my new favourite test: draw a septagon using only ASCII. DeepSeek and ChatGPT both fail so badly...
Long shot, given the limited reach so far here and the late stage in the process, but if you have (or are) a student that may be interested, please reach out!
January 25, 2025 at 3:20 PM
Long shot, given the limited reach so far here and the late stage in the process, but if you have (or are) a student that may be interested, please reach out!
This year, the proof of A* optimality in our AI class happens to be scheduled for a wintery 8:30am lecture. The random few who make it and tune in will be legendary.
January 23, 2025 at 1:20 PM
This year, the proof of A* optimality in our AI class happens to be scheduled for a wintery 8:30am lecture. The random few who make it and tune in will be legendary.
Reposted by Christian Muise
glad to see we invented Deep Gaslighting
January 22, 2025 at 3:00 PM
glad to see we invented Deep Gaslighting
Reposted by Christian Muise
A thoughtful piece on AI and education from @emollick.bsky.social
"The integration of AI in education is not a future possibility—it's our present reality... It requires a fundamental reimagining of how we teach, learn, and assess knowledge."
www.oneusefulthing.org/p/post-apoca...
"The integration of AI in education is not a future possibility—it's our present reality... It requires a fundamental reimagining of how we teach, learn, and assess knowledge."
www.oneusefulthing.org/p/post-apoca...
Post-apocalyptic education
What comes after the Homework Apocalypse
www.oneusefulthing.org
January 12, 2025 at 12:58 AM
A thoughtful piece on AI and education from @emollick.bsky.social
"The integration of AI in education is not a future possibility—it's our present reality... It requires a fundamental reimagining of how we teach, learn, and assess knowledge."
www.oneusefulthing.org/p/post-apoca...
"The integration of AI in education is not a future possibility—it's our present reality... It requires a fundamental reimagining of how we teach, learn, and assess knowledge."
www.oneusefulthing.org/p/post-apoca...