Andrew Cooper
@pocooperose.bsky.social
Associate Professor, Teaching Stream @UTSC Psych & GD-PCS 🇨🇦 Opinions my own. (he/him)
Therapy mechanisms & preferences (CBT, MDD, PTSD, dropout); meta-analysis & meta-science; hidden curriculum & sci comm.
Lately: psychology & videogames 🎮👾
Therapy mechanisms & preferences (CBT, MDD, PTSD, dropout); meta-analysis & meta-science; hidden curriculum & sci comm.
Lately: psychology & videogames 🎮👾
Pinned
Pinning this for video game makers, media, players, scholars, etc who I’m following on Bluesky —
I’m launching a new course called *Psychology and Videogames* in 2026 & watching all the awesome stuff you folks are working on for inspiration! (On top of what I follow/watch/listen to for fun :)
I’m launching a new course called *Psychology and Videogames* in 2026 & watching all the awesome stuff you folks are working on for inspiration! (On top of what I follow/watch/listen to for fun :)
Reposted by Andrew Cooper
i got this killing a necromancer in elden ring
December 5, 2025 at 7:11 PM
i got this killing a necromancer in elden ring
Reposted by Andrew Cooper
I am a suicide attempt survivor so I’m kind of selfishly interested in prevention and, again, while it may difficult to completely prevent IDEATION, what ChatGPT does (despite guardrails!) is *cheerlead and facilitate* completion. You can come back from ideation but not a chat-optimized completion.
This is not true. This is argument that pro-gun people use to discount red flag and waiting period laws. Ease of access and lethality (let alone cheerful instruction in the best method!) is the #1 predictor of attempts leading actual completion. The more barriers, the more people live.
I think it's fine Chat GPT is telling people how to kill themselves. They'll figure it out one way or another if they're serious.
December 4, 2025 at 4:57 PM
I am a suicide attempt survivor so I’m kind of selfishly interested in prevention and, again, while it may difficult to completely prevent IDEATION, what ChatGPT does (despite guardrails!) is *cheerlead and facilitate* completion. You can come back from ideation but not a chat-optimized completion.
Reposted by Andrew Cooper
it's wild how the positive use cases are like "save 5 minutes reading a long doc" and "write a generic email faster" and the negative use cases are "get encouraged to dive deeper in psychosis, misogyny, or suicidal thoughts" and we're still on the fence about overall value
New from 404 Media: ChatGPT told a violent stalker to continue running his misogynistic and threat-filled podcast despite the “haters,” and to visit more gyms to find women. The man stalked women in multiple states, including touching and following from work
www.404media.co/chatgpt-spot...
www.404media.co/chatgpt-spot...
ChatGPT Told a Violent Stalker to Embrace the 'Haters,' Indictment Says
A newly filed indictment claims a wannabe influencer used ChatGPT as his "therapist" and "best friend" in his pursuit of the "wife type," while harassing women so aggressively they had to miss work a...
www.404media.co
December 3, 2025 at 6:17 PM
it's wild how the positive use cases are like "save 5 minutes reading a long doc" and "write a generic email faster" and the negative use cases are "get encouraged to dive deeper in psychosis, misogyny, or suicidal thoughts" and we're still on the fence about overall value
Reposted by Andrew Cooper
important on LLMs for academics:
1️⃣ LLMs are usefully seen as lossy content-addressable systems
2️⃣ we can't automatically detect plagiarism
3️⃣ LLMs automate plagiarism & paper mills
4️⃣ we must protect literature from pollution
5️⃣ LLM use is a CoI
6️⃣ prompts do not cause output in authorial sense
1️⃣ LLMs are usefully seen as lossy content-addressable systems
2️⃣ we can't automatically detect plagiarism
3️⃣ LLMs automate plagiarism & paper mills
4️⃣ we must protect literature from pollution
5️⃣ LLM use is a CoI
6️⃣ prompts do not cause output in authorial sense
![5 Ghostwriter in the Machine
A unique selling point of these systems is conversing and writing in a human-like way. This is imminently understandable, although wrong-headed, when one realises these are systems that
essentially function as lossy2
content-addressable memory: when
input is given, the output generated by the model is text that
stochastically matches the input text. The reason text at the output looks novel is because by design the AI product performs
an automated version of what is known as mosaic or patchwork
plagiarism (Baždarić, 2013) — due to the nature of input masking and next token prediction, the output essentially uses similar words in similar orders to what it has been exposed to. This
makes the automated flagging of plagiarism unlikely, which is
also true when students or colleagues perform this type of copypaste and then thesaurus trick, and true when so-called AI plagiarism detectors falsely claim to detect AI-produced text (Edwards, 2023a). This aspect of LLM-based AI products can be
seen as an automation of plagiarism and especially of the research paper mill (Guest, 2025; Guest, Suarez, et al., 2025; van
Rooij, 2022): the “churn[ing] out [of] fake or poor-quality journal papers” (Sanderson, 2024; Committee on Publication Ethics,](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:vsyj5jtayxcq5dw5mtja4xtb/bafkreieufyngjbrn23zkmesrimcwemxkxtaf2x7d3bedjt7bg65htgtkrm@jpeg)
October 10, 2025 at 10:33 AM
important on LLMs for academics:
1️⃣ LLMs are usefully seen as lossy content-addressable systems
2️⃣ we can't automatically detect plagiarism
3️⃣ LLMs automate plagiarism & paper mills
4️⃣ we must protect literature from pollution
5️⃣ LLM use is a CoI
6️⃣ prompts do not cause output in authorial sense
1️⃣ LLMs are usefully seen as lossy content-addressable systems
2️⃣ we can't automatically detect plagiarism
3️⃣ LLMs automate plagiarism & paper mills
4️⃣ we must protect literature from pollution
5️⃣ LLM use is a CoI
6️⃣ prompts do not cause output in authorial sense
Votes are in: they're all bad, but UBC wins "Worst Reference Letter Submission Portal" of the 2025-2026 season. Comment boxes under every ranking box. Stalls out just trying to load. Crashed multiple times. Erased typed responses. Truly inane instructions. Disappeared the "UPLOAD FILES" button, 2x.
December 1, 2025 at 5:50 PM
Votes are in: they're all bad, but UBC wins "Worst Reference Letter Submission Portal" of the 2025-2026 season. Comment boxes under every ranking box. Stalls out just trying to load. Crashed multiple times. Erased typed responses. Truly inane instructions. Disappeared the "UPLOAD FILES" button, 2x.
Reposted by Andrew Cooper
This is as good as this stuff will ever be, this is the phase where we're all supposed to be getting hooked in to this disruptive new thing. Once the free money dries up and they have to turn a profit, it will inevitably get worse and worse
"All of this falls apart if humans don't adopt the tech. This is why you've seen Meta cram its lame chatbots into WhatsApp and Instagram. This is why Notepad and Paint now have useless Copilot buttons on Windows. This is why Google Gemini wants to "help you" read and reply to your emails."
Analysis: OpenAI is a loss-making machine, how can it survive?
Don't call it a bubble! Loss-making monster OpenAI is on the hook for $1.4 trillion (with a T) in compute commitments. How can this go on?
www.windowscentral.com
November 30, 2025 at 4:31 PM
This is as good as this stuff will ever be, this is the phase where we're all supposed to be getting hooked in to this disruptive new thing. Once the free money dries up and they have to turn a profit, it will inevitably get worse and worse
Reposted by Andrew Cooper
I think about this tweet a lot
November 30, 2025 at 4:23 AM
I think about this tweet a lot
Reposted by Andrew Cooper
What a great headline
November 28, 2025 at 3:49 PM
What a great headline
Reposted by Andrew Cooper
i know WHY this is but that just makes it even more of a kick in the pants lmao
November 28, 2025 at 3:28 AM
i know WHY this is but that just makes it even more of a kick in the pants lmao
Reposted by Andrew Cooper
The enshitification of search in one image.
November 26, 2025 at 1:33 PM
The enshitification of search in one image.
Reposted by Andrew Cooper
"It is against the TOS to be talked into suicide by our LLM" is possibly the most evil tech bro thing I've ever read
Additionally, OpenAI argues its not liable because Raine, by using ChatGPT for self-harm, broke its terms of service
November 26, 2025 at 12:29 AM
"It is against the TOS to be talked into suicide by our LLM" is possibly the most evil tech bro thing I've ever read
Reposted by Andrew Cooper
Look, LLM use in academia isn't just students. You need to be speaking clearly with your collaborators about your writing standards from the beginning. Make people say out loud that they don't write their own articles anymore.
November 25, 2025 at 12:41 PM
Look, LLM use in academia isn't just students. You need to be speaking clearly with your collaborators about your writing standards from the beginning. Make people say out loud that they don't write their own articles anymore.
Reposted by Andrew Cooper
Relying on ChatGPT to teach you about a topic leaves you with shallower knowledge than Googling and reading about it, according to new research that compared what more than 10,000 people knew after using one method or the other.
Shared by @gizmodo.com: buff.ly/yAAHtHq
Shared by @gizmodo.com: buff.ly/yAAHtHq
November 21, 2025 at 11:48 AM
Relying on ChatGPT to teach you about a topic leaves you with shallower knowledge than Googling and reading about it, according to new research that compared what more than 10,000 people knew after using one method or the other.
Shared by @gizmodo.com: buff.ly/yAAHtHq
Shared by @gizmodo.com: buff.ly/yAAHtHq
Bards proving themselves indispensable members of the party once again…
absolutely LOVE this preprint on tricking LLMs using poetry:
arxiv.org/abs/2511.15304
Like magic spells, you need to use the correct pentameter or the demon won't listen
arxiv.org/abs/2511.15304
Like magic spells, you need to use the correct pentameter or the demon won't listen
Adversarial Poetry as a Universal Single-Turn Jailbreak Mechanism in Large Language Models
We present evidence that adversarial poetry functions as a universal single-turn jailbreak technique for Large Language Models (LLMs). Across 25 frontier proprietary and open-weight models, curated po...
arxiv.org
November 21, 2025 at 9:52 AM
Bards proving themselves indispensable members of the party once again…
Reposted by Andrew Cooper
this is the correct framing
Republicans vote overwhelmingly to pretend they want the Epstein Files released. www.axios.com/2025/11/18/h...
House votes to force DOJ release of Jeffrey Epstein files
While the vote marks an important step in backers' crusade to get the files released, significant barriers remain.
www.axios.com
November 18, 2025 at 7:55 PM
this is the correct framing
Reposted by Andrew Cooper
Like it or not this technology is here to stay and if you don’t use it you’ll fall behind in your industry
November 16, 2025 at 11:49 PM
Like it or not this technology is here to stay and if you don’t use it you’ll fall behind in your industry
Reposted by Andrew Cooper
Ireland winning goal and end of match with Irish commentary
November 16, 2025 at 4:30 PM
Ireland winning goal and end of match with Irish commentary
Reposted by Andrew Cooper
Outside of Motorsport, this is easily one of my favourite ever sporting moments. A petulant child getting his comeuppance, on the pitch at least. Even sweeter when it's against my country 😹
🎥 RTÉ Sport
🎥 RTÉ Sport
November 14, 2025 at 6:08 AM
Outside of Motorsport, this is easily one of my favourite ever sporting moments. A petulant child getting his comeuppance, on the pitch at least. Even sweeter when it's against my country 😹
🎥 RTÉ Sport
🎥 RTÉ Sport
Reposted by Andrew Cooper
The link between the gut #microbiome and autism is not backed by science, researchers say.
Read the full opinion piece in @cp-neuron.bsky.social: spkl.io/63322AbxpA
@wiringthebrain.bsky.social, @statsepi.bsky.social, & @deevybee.bsky.social
Read the full opinion piece in @cp-neuron.bsky.social: spkl.io/63322AbxpA
@wiringthebrain.bsky.social, @statsepi.bsky.social, & @deevybee.bsky.social
November 13, 2025 at 4:00 PM
The link between the gut #microbiome and autism is not backed by science, researchers say.
Read the full opinion piece in @cp-neuron.bsky.social: spkl.io/63322AbxpA
@wiringthebrain.bsky.social, @statsepi.bsky.social, & @deevybee.bsky.social
Read the full opinion piece in @cp-neuron.bsky.social: spkl.io/63322AbxpA
@wiringthebrain.bsky.social, @statsepi.bsky.social, & @deevybee.bsky.social
Reposted by Andrew Cooper
early career scholar and their first publication
November 9, 2025 at 9:41 AM
early career scholar and their first publication
Reposted by Andrew Cooper
Six people died from tampered Tylenol and the company pulled 30 million bottles off the market. ChatGPT is accused of urging seven people towards suicide, and OpenAI just assures us they’re still working out the kinks. Billions in investment, zero accountability.
OpenAI faces 7 lawsuits claiming ChatGPT drove people to suicide, delusions
OpenAI is facing seven lawsuits claiming ChatGPT drove people to suicide and harmful delusions even when they had no prior mental health issues.
apnews.com
November 8, 2025 at 2:46 PM
Six people died from tampered Tylenol and the company pulled 30 million bottles off the market. ChatGPT is accused of urging seven people towards suicide, and OpenAI just assures us they’re still working out the kinks. Billions in investment, zero accountability.
Reposted by Andrew Cooper
OpenAI pirated large numbers of books and used them to train models.
OpenAI then deleted the dataset with the pirated books, and employees sent each other messages about doing so.
A lawsuit could now force the company to pay $150,000 per book, adding up to billions in damages.
OpenAI then deleted the dataset with the pirated books, and employees sent each other messages about doing so.
A lawsuit could now force the company to pay $150,000 per book, adding up to billions in damages.
November 4, 2025 at 3:30 PM
OpenAI pirated large numbers of books and used them to train models.
OpenAI then deleted the dataset with the pirated books, and employees sent each other messages about doing so.
A lawsuit could now force the company to pay $150,000 per book, adding up to billions in damages.
OpenAI then deleted the dataset with the pirated books, and employees sent each other messages about doing so.
A lawsuit could now force the company to pay $150,000 per book, adding up to billions in damages.
Reposted by Andrew Cooper
The problem with the Torment Nexus is that it wasn’t optimized for user experience. Average daily users experience a ratio of roughly 70% torment vs non-torment in the nexus. The changes we’re implementing should reduce that to a staggering 60% torment. Our goal is 50/50 torment/non-torment by 2028.
November 1, 2025 at 5:14 PM
The problem with the Torment Nexus is that it wasn’t optimized for user experience. Average daily users experience a ratio of roughly 70% torment vs non-torment in the nexus. The changes we’re implementing should reduce that to a staggering 60% torment. Our goal is 50/50 torment/non-torment by 2028.
Reposted by Andrew Cooper
If the federal Liberals use generative AI in anything having to do with serious governing, this automatically becomes the stupidest government of my lifetime
If the Carney government deploys generative AI -- a product that by design cannot be trusted to provide reliable information -- to answer Canadians' tax question, it should be required to forgive the taxes of anyone who follows any incorrect answers.
Tax agency looking at AI, training to help call centre staff provide accurate answers
OTTAWA - A senior official working at the Canada Revenue Agency says the agency is looking to artificial intelligence and better training to help call centre agents provide more accurate
www.thestar.com
October 29, 2025 at 1:14 AM
If the federal Liberals use generative AI in anything having to do with serious governing, this automatically becomes the stupidest government of my lifetime