Michael Black
@mblack.us
Reposted by Michael Black
much of the work on AI-generated misinformation focuses on regular users/communities, while ignoring how political and economic elites can seal themselves into a hermetic slop box -- a repeat of the same mistake we made with radicalization and conspiracism in the previous decade
oh this is absolutely what i think. i think they feed him fake polls and outright AI slop
At this point, I wonder if he isn't lying when he says he has "my highest Poll Numbers, ever" since it's entirely possible that's what his aides are telling him. It's quite likely that there's an entire Potemkin village built around Trump right now.
November 26, 2025 at 2:49 PM
much of the work on AI-generated misinformation focuses on regular users/communities, while ignoring how political and economic elites can seal themselves into a hermetic slop box -- a repeat of the same mistake we made with radicalization and conspiracism in the previous decade
Reposted by Michael Black
"We should measure the impact of actually existing AI rather than worry about terminators or gray goo. And in this respect, while it's early days yet, the labor market effects are mostly justifying layoffs rather than increasing productivity. But there are strong use cases in surveillance fascism."
a painting of a man wearing sunglasses and a scarf with stars on it
ALT: a painting of a man wearing sunglasses and a scarf with stars on it
media.tenor.com
November 26, 2025 at 3:14 PM
"We should measure the impact of actually existing AI rather than worry about terminators or gray goo. And in this respect, while it's early days yet, the labor market effects are mostly justifying layoffs rather than increasing productivity. But there are strong use cases in surveillance fascism."
I would echo this for writing instruction, specifically the idea that students should just revise/edit LLM output.
I will add the following: our students lack the research skills required to audit an LLM essay for errors. They don’t arrive on campus with these skills; we teach it to them over four long years. So throwing freshmen in the deep end and saying “swim your way to a shore of rectitude” is folly.
November 25, 2025 at 1:29 AM
I would echo this for writing instruction, specifically the idea that students should just revise/edit LLM output.
Reposted by Michael Black
In some cases, maybe. But this exercise often won’t work because students (like most of us) are very bad at intuitively understanding the limits of their knowledge without being trained to recognize it, and “fact check the AI” quickly becomes “using AI to fact check AI” and the cycle repeats.
November 24, 2025 at 1:17 PM
In some cases, maybe. But this exercise often won’t work because students (like most of us) are very bad at intuitively understanding the limits of their knowledge without being trained to recognize it, and “fact check the AI” quickly becomes “using AI to fact check AI” and the cycle repeats.
Reposted by Michael Black
Asking a chatbot isn't better than nothing -- it's worse than nothing. Unless the goal isn't learning but rather "turning in assignments".
November 24, 2025 at 2:51 AM
Asking a chatbot isn't better than nothing -- it's worse than nothing. Unless the goal isn't learning but rather "turning in assignments".
Reposted by Michael Black
There is a flaw in the argument, “This is the worst LLMs will ever be. They will get better.”
Consider: when people pay for token usage, there is an incentive for building bad models to maximize token usage. The models will not necessarily get better but instead be designed maximize user payment.
Consider: when people pay for token usage, there is an incentive for building bad models to maximize token usage. The models will not necessarily get better but instead be designed maximize user payment.
November 24, 2025 at 12:47 AM
There is a flaw in the argument, “This is the worst LLMs will ever be. They will get better.”
Consider: when people pay for token usage, there is an incentive for building bad models to maximize token usage. The models will not necessarily get better but instead be designed maximize user payment.
Consider: when people pay for token usage, there is an incentive for building bad models to maximize token usage. The models will not necessarily get better but instead be designed maximize user payment.
Reposted by Michael Black
Reposted by Michael Black
Based on our research (although this not mentioned cwa-union.org/ghost-worker...) and featuring data workers you should follow here including @kristap.bsky.social & @brookh.bsky.social!!
November 22, 2025 at 5:16 PM
Based on our research (although this not mentioned cwa-union.org/ghost-worker...) and featuring data workers you should follow here including @kristap.bsky.social & @brookh.bsky.social!!
Reposted by Michael Black
Every time I see a study trying to use AI to improve early childhood development it seems designed as a desperate wish for AI to be useful to a field where it has no practical application and likely never will. Secondarily, a grasp for funding in the over-funded shiny object of optimizing childhood.
“This study is trying to understand how we can leverage AI and machine-learning principles to improve recognition for toddlers and preschoolers.”
Canadian researchers look into how AI can better understand toddler talk like 'doodidees' | CBC News
Two researchers at Western University in London, Ont., are studying how to improve the way artificial intelligence understands toddlers' speaking patterns. They're recruiting preschool-age children to...
www.cbc.ca
November 22, 2025 at 5:18 PM
Every time I see a study trying to use AI to improve early childhood development it seems designed as a desperate wish for AI to be useful to a field where it has no practical application and likely never will. Secondarily, a grasp for funding in the over-funded shiny object of optimizing childhood.
Reposted by Michael Black
everyone on a game is a storyteller and the storytellers on Space Warlord Baby Trading Simulator fuckin’ despise the incentives of late stage capitalism
also that kid beto is pretty cool
also that kid beto is pretty cool
November 22, 2025 at 1:44 AM
everyone on a game is a storyteller and the storytellers on Space Warlord Baby Trading Simulator fuckin’ despise the incentives of late stage capitalism
also that kid beto is pretty cool
also that kid beto is pretty cool
The spirit is willing but my retirement accounts are weak.
this is like one of those online quizzes where 90% of people on earth will come away saying "yep, that's me"
November 21, 2025 at 3:53 PM
The spirit is willing but my retirement accounts are weak.
Reposted by Michael Black
talkin' about the AI bubble
November 20, 2025 at 6:56 PM
talkin' about the AI bubble
Reposted by Michael Black
Instead of trying out the teacher version of ChatGPT or worrying about the AI bubble, I'd love for you to read my 4,500 word (lol, what?!) blog post about AI and genre. I poured myself into it, literally.
meresophistry.substack.com/p/the-generi...
meresophistry.substack.com/p/the-generi...
The generic abyss of artificial intelligence
A long, long, long blog post about AI and genre
meresophistry.substack.com
November 20, 2025 at 2:56 PM
Instead of trying out the teacher version of ChatGPT or worrying about the AI bubble, I'd love for you to read my 4,500 word (lol, what?!) blog post about AI and genre. I poured myself into it, literally.
meresophistry.substack.com/p/the-generi...
meresophistry.substack.com/p/the-generi...
Reposted by Michael Black
If you use GMail, AI (Gemini) was turned on yesterday by default and now scans all of your content for machine learning. To turn off, go to Settings>General and scroll down. Uncheck the box for "Smart features."
There's other "Smart" add-ons as well, but that's the one that reads your content.
There's other "Smart" add-ons as well, but that's the one that reads your content.
November 20, 2025 at 5:32 PM
If you use GMail, AI (Gemini) was turned on yesterday by default and now scans all of your content for machine learning. To turn off, go to Settings>General and scroll down. Uncheck the box for "Smart features."
There's other "Smart" add-ons as well, but that's the one that reads your content.
There's other "Smart" add-ons as well, but that's the one that reads your content.
Reposted by Michael Black
A little belatedly sharing this piece that I published in @chronicle.com 's Opinion Forum (sorry full text is paywalled)
www.chronicle.com/article/how-...
www.chronicle.com/article/how-...
November 20, 2025 at 12:46 AM
A little belatedly sharing this piece that I published in @chronicle.com 's Opinion Forum (sorry full text is paywalled)
www.chronicle.com/article/how-...
www.chronicle.com/article/how-...
Reposted by Michael Black
I don’t know who needs to hear this, but AP exams won’t fix the literacy crisis either.
November 19, 2025 at 3:45 PM
I don’t know who needs to hear this, but AP exams won’t fix the literacy crisis either.
Starting to wonder if stuff like this is why I mostly play history themed spreadsheet simulators these days.
November 18, 2025 at 2:46 AM
Starting to wonder if stuff like this is why I mostly play history themed spreadsheet simulators these days.
Reposted by Michael Black
What this reveals is that for most people in charge right now “AI” is less useful as a technology than as a piece of language to shift a conversation in whatever direction they need it to go. In one minute it will propel a new economy; in another it’s the reason for a recession.
November 17, 2025 at 1:39 PM
What this reveals is that for most people in charge right now “AI” is less useful as a technology than as a piece of language to shift a conversation in whatever direction they need it to go. In one minute it will propel a new economy; in another it’s the reason for a recession.
Reposted by Michael Black
This below is in the context of computer science but we all need to be having this conversation about whether "using AI" in classes prepares you for future jobs or whether actually doing the thing yourself prepares you for a world in which you may or may not use AI. The idea that using AI to write/1
![A degree in CS mostly prepares students to become computer scientists, just as a history degree mostly prepares students to be historians. The idea is not that every student will become an academic— few do—but that the initial training in this direction will prepare students for varied pursuits. When the world changes, Erat argues, the student who studied CS will have the creativity to “learn how to search, how to integrate things.” But for the student who just used AI to make programs, he said, “It’s going to be impossible.”
Conversely, CS professor Lin Zhong thinks Yale’s core CS curriculum is “stuck in the 1980s.” He wants students to develop the AI expertise that they would need as professional software engineers. “If your student[s] don’t learn how to use [AI], they lose their job,” he said. Students in Zhong’s Computer Systems Design course earn extra credit for using AI creatively to do their homework.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:a6m5puxjmwuwxhuviabkhg67/bafkreidr3hlf3v5bi3c4dwzguxvi2khyjm4tmr5unm4funrdwyqjgvys2m@jpeg)
November 16, 2025 at 5:24 PM
This below is in the context of computer science but we all need to be having this conversation about whether "using AI" in classes prepares you for future jobs or whether actually doing the thing yourself prepares you for a world in which you may or may not use AI. The idea that using AI to write/1
But [Google engineer in Education for AI] Phal wasn’t sure how this would work in practice. “I’m not an expert on how to teach,” he said. Phal thinks professors should view the potential of teaching with AI as “something transformative." 🙄🙄🙄
Inside Yale’s Quiet Reckoning with AI | The New Journal
Amid ChatGPT's rising popularity and a computer science cheating scandal, Yale students, professors, and administrators wrestle privately with the proper role of AI in education. What happens when eve...
thenewjournalatyale.com
November 17, 2025 at 1:14 PM
But [Google engineer in Education for AI] Phal wasn’t sure how this would work in practice. “I’m not an expert on how to teach,” he said. Phal thinks professors should view the potential of teaching with AI as “something transformative." 🙄🙄🙄
Reposted by Michael Black
Non-industry compromised scientists keep saying these models don't become safe, no matter what, but people keep thinking just because the concept of guardrails is mentioned it must work. By definition, it doesn't. This is not something open to discussion, unless you're a paid shill.
"Powered by OpenAI’s GPT-4o model by default...tests repeatedly showed that the AI toy dropped its guardrails the longer a conversation went on, until hitting rock bottom on incredibly disturbing topics."
AI-Powered Stuffed Animal Pulled From Market After Disturbing Interactions With Children
FoloToy says it's suspended sales of its AI-powered teddy bear after researchers found it gave wildly inappropriate and dangerous answers.
futurism.com
November 17, 2025 at 5:18 AM
Non-industry compromised scientists keep saying these models don't become safe, no matter what, but people keep thinking just because the concept of guardrails is mentioned it must work. By definition, it doesn't. This is not something open to discussion, unless you're a paid shill.
Reposted by Michael Black
This site brilliantly demonstrates the perils of relying on LLM generated content. It creates code to show a clock from nine popular models every minute. Changes within each model are alarming. Keep it on screen for 10 minutes, and tell me how you feel about relying on LLMs! clocks.brianmoore.com
AI World Clocks
The current time as rendered by 9 different AI models. By Brian Moore.
clocks.brianmoore.com
November 14, 2025 at 10:02 PM
This site brilliantly demonstrates the perils of relying on LLM generated content. It creates code to show a clock from nine popular models every minute. Changes within each model are alarming. Keep it on screen for 10 minutes, and tell me how you feel about relying on LLMs! clocks.brianmoore.com
Reposted by Michael Black
here_we_go_again.webp
November 14, 2025 at 2:58 PM
here_we_go_again.webp
Reposted by Michael Black
Keep the college essay.
You don't have to fulfill the prophecy that the college essay is dead because some jaded dudes in the Atlantic, New Yorker, and NYT told you so two years ago.
It's OK. You can keep the essay.
You don't have to fulfill the prophecy that the college essay is dead because some jaded dudes in the Atlantic, New Yorker, and NYT told you so two years ago.
It's OK. You can keep the essay.
November 14, 2025 at 2:47 PM
Keep the college essay.
You don't have to fulfill the prophecy that the college essay is dead because some jaded dudes in the Atlantic, New Yorker, and NYT told you so two years ago.
It's OK. You can keep the essay.
You don't have to fulfill the prophecy that the college essay is dead because some jaded dudes in the Atlantic, New Yorker, and NYT told you so two years ago.
It's OK. You can keep the essay.
Reposted by Michael Black
Allowing for the refusal of AI in our educational institutions is simply the right thing to do. If we can's support those freedoms, what are we doing? Congratulations on a year of vital work.
A new post on Refusing GenAI in Writing Studies from @jsanofranchini.bsky.social, @rcmeg.bsky.social, and me: "A Year of Refusal"
refusinggenai.wordpress.com/2025/11/12/a...
refusinggenai.wordpress.com/2025/11/12/a...
A Year of Refusal
Maggie Fernandes, University of ArkansasMegan McIntyre, University of ArkansasJennifer Sano-Franchini, West Virginia University Photo by Francis Glenn Marciano on Pexels.com It’s been about a year …
refusinggenai.wordpress.com
November 13, 2025 at 3:02 PM
Allowing for the refusal of AI in our educational institutions is simply the right thing to do. If we can's support those freedoms, what are we doing? Congratulations on a year of vital work.