Large language models are language machines, and we need to think about what they do with language as more than just "mathy math." Language Machines as Antimeme www.ailog.blog/p/language-m...

@orzelc.bsky.social has become one of my favorite higher education writers of late. But he is importantly wrong when he disagrees with Timothy Burke’s description of “the deal” offered to nine universities. Trump offers carrots to nine universities www.ailog.blog/p/trump-offe...

Why were so many university presidents and boards quick to undermine the principles and practice of academic freedom this past week? The answer has to do with who runs institutions of higher education, and how they think and don’t think about their work. www.ailog.blog/p/its-not-ch...

This essay is better than the talk I gave at AC&U 2025-26 Institute on AI, Pedagogy, and the Curriculum on Friday because it reflects the collective intelligence of those who asked questions and made comments. What comes after the chatbot? www.ailog.blog/p/what-comes...

The Lippmann-Dewey debate is a meme created in the 1980s, and used ever since to talk about a question intellectuals find endlessly fascinating: wouldn’t it be better if intellectuals ran things? Lippmann called this “as complete a delusion as perpetual motion.” www.ailog.blog/p/a-revery-e...

Nicholas Carr’s Superbloom brings to light our longstanding cultural habit of enthusiastically greeting each new social technology. You may be surprised to learn that this cycle goes back a ways. With a Certain Pathetic Moderation: AI as a New Social Medium www.ailog.blog/p/with-a-cer...

Ben Recht says it clearly and plainly. The way this technology is being commercialized matters. OpenAI is the worst. Courts should punish them. Consumers should boycott them. No educational institution should do business with them. The Banal Evil of AI Safety www.argmin.net/p/the-banal-...

𝐀𝐈 𝐋𝐨𝐠 turns two years old tomorrow. To celebrate, I made a listicle and decided to reveal Phase 2 of my plan. Five things I have learned writing AI Log + Phase 2 begins www.ailog.blog/p/five-thing...

I will teach a class enrolling only first-year undergraduates this fall for the first time in years. We're going to have some fun reading, writing, and talking about AI. Shakespeare helped me figure out how to start things off. Get Me to the Funnery www.ailog.blog/p/get-me-to-...

here's part 2 of the conversation with @ailogblog.bsky.social & the conversations he's looking to have with #highereducation institutions about #GenAI aiedusimplified.substack.com/p/give-ourse... #AIEdu #EdTech

Part 1 of a conversation with the great, @ailogblog.bsky.social about what he's doing in the classroom around #GenAI #HigherEd #AIEdu #EdTech #Teaching aiedusimplified.substack.com/p/in-the-mod...

This is not the first time newly elected Republicans have interested themselves in the affairs of universities and colleges. In 1819, we got corporate personhood redefined. What will get we get this time? To what problem is the modern American university a solution? www.ailog.blog/p/to-what-pr...

"A university is definitely not a democratic institution" is truer today than when Dean Herbert Deane said it in 1968, and more true of Columbia than it was a week ago. Three bad ideas that are good for US higher education www.ailog.blog/p/three-bad-...

Small matters www.ailog.blog/p/small-matt...

Socrates comes to mind when I am urged to teach AI literacy. There is a rumor, likely false, that Socrates himself was illiterate. Maybe he chose not to read and write, so great was his commitment to the kind of conversational back and forth we named after him. www.ailog.blog/p/a-phaedrus...

Brad DeLong posted a double shot of analysis of what he calls Modern Advanced Machine Learning Models (MAMLMs). An excellent overview of the economic and organizational impact of this new cultural technology. Spreadsheet, Not Skynet: Microdoses, Not Microprocessors open.substack.com/pub/braddelo...

A defining characteristic of large AI models is to spiral or zigzag, introducing errors that no human would make. It is a better Clippy, a better tutor, and a better Eliza, except when it isn’t. After the AI bubble: ChatGPT as an off-modern educational technology www.ailog.blog/p/after-the-...

In my excitement to get to Q & A, I forgot to share the link to the references and image credits at a talk I gave earlier this week. Since several attendees started following me here, I’m sharing it now. Notes and image credits for "The Boring Revolution" www.ailog.blog/p/notes-and-...

Do large language models bring anything of value to the classroom or the workplace? If we stop pretending they are human, and instead, treat them as a new cultural technology, we might find out. Seriously, Let's Stop Treating LLMs like People www.ailog.blog/p/seriously-...

Wonderful piece illustrating how AI is a "normal technology" that has been marketed and used in ways that imagine it as something much, much more. I got fooled by AI-for-science hype—here's what it taught me www.understandingai.org/p/i-got-fool...

Teaching with AI Redux www.ailog.blog/p/teaching-w...

Instead of almost AGI or a cultural apocalypse, let's think of large AI models as: a cultural and social technology a normal technology a technology to be used in the revival of management cybernetics. Here is a list. On Beyond AGI: A reading list of sorts www.ailog.blog/p/on-beyond-...

Does Ben Breen contradict himself? I find this essay large, to borrow Whitman’s famous line, and it waits on the door slab for the rest of us. AI makes the humanities more important, but also a lot weirder open.substack.com/pub/resobscu...

On Monday, @benjaminjriley.bsky.social said mean things about NGO leaders and bureaucrats who celebrated Trump's executive order on AI in education. It made some people sad and confused. Kenneth Burke helps explain what's going on. Nobody here but us chickens www.ailog.blog/p/nobody-her...

Large AI models are mycelia. Their outputs are occasionally delightful, often strange, and in some cases, unsafe to consume. There is a mushroom in my computer www.ailog.blog/p/there-is-a...

Scott Alexander researches Stafford Beer by reading Twitter and reports that Beer is a lightweight. I’m with Kevin Munger on this one: “I'm glad that Stafford Beer is getting famous enough to be vacuously quoted by random people on Twitter!” www.ailog.blog/p/old-ways-o...

Big news in the let’s-develop-alternatives-to-AGI-instead-of-endlessly-arguing-about-AGI front! One way to start is to talk about them as normal technology rather than as magic that will replace or destroy humans. AI as Normal Technology www.aisnakeoil.com/p/ai-as-norm...

Anthropic reports on education and announced "partnerships" with research universities, the answer to OpenAI's NextGenAI consortia. If OpenAI and Anthropic offered to pay you to be their customer, which would you choose? Boring revolutions and large-elbowed models www.ailog.blog/p/boring-rev...

The endless back and forth over AGI is getting stale. We need better ways to talk about what people should do when they use large AI models. @alisongopnik.bsky.social @himself.bsky.social @dsquareddigest.bsky.social offer that and more. On Beyond AGI! www.ailog.blog/p/on-beyond-...

The important question facing teachers is which digital tools to use and why. I think the answer in class is zero because class time needs to focus on something besides digital tools. What is an LLM doing in my classroom? Part 4: What's Next www.ailog.blog/p/what-is-an...

At the end of this entertaining essay about AGI, Benjamin Riley reveals he will be speaking at the ASU+GSV Summit on "AI Will Not Revolutionize Education." I can't think of a title that cuts so directly against the grain of this conference. open.substack.com/pub/buildcog...

Arresting students, threatening enforcement actions, and canceling grants and contracts are causing chaos on campuses across the US. What happened to NYC last month means it could get worse. Can the Trump-Musk regime claw back funds already dispersed to universities? www.ailog.blog/p/can-the-tr...

John Warner seems to have made it into the small club that actually makes a living as a writer. Will that club still exist next year? My answer is "yes," and his explanation of the value of writing is a big part of the reason. Writing after the homework apocalypse www.ailog.blog/p/writing-af...

I have frequently borrowed Kevin Drum's charts. His passing is a loss for those who love blogging. This is a wonderful tribute to the work he did as a writer. And a lesson about how to think about data. A remembrance of Kevin Drum and his time-series charts open.substack.com/pub/computin...

Like the press release last month announcing its partnership with California State University, I see NextGenAI as an act of desperation by a company that has yet to figure out a profitable strategy for selling what it hopes turns out to be an educational technology. www.ailog.blog/p/openais-ne...

The hype about how AI will create efficiencies combined with hiring and admissions freezes is a giant trap hidden in plain sight amid the turmoil and chaos of this month. It's Bad open.substack.com/pub/ailogblo...

There is never a good time to make bad decisions, but as the impact of what’s happening in Washington is felt on campuses throughout the country, we should call out wasteful spending. Home fires burning open.substack.com/pub/ailogblo...

Maybe thinking of LLMs as confabulators can help us understand what generative AI is actually good for. On Confabulation open.substack.com/pub/ailogblo...

CSU's announcement that everyone gets a ChatGPT subscription has become more noise in AI Fight Club. The signal I was trying to get through is that OpenAI really is the worst option available. Anthropic made that point yesterday with its new Economic Index. substack.com/@erob/note/c...

EdSource quoted a CSU official saying the system will pay “$16.9 million over the lifetime of its partnership with OpenAI.” That is real money to CSU, given the cuts to higher ed in the current state education budget. For OpenAI, it is a drop in their leaky bucket. substack.com/@erob/note/c...

CSU's announcement on Monday sets a precedent for treating OpenAI as an established educational technology company. Really it is a tech start-up run by a CEO with questionable ethics who accidentally created a hit consumer product based on new chatbot technology. open.substack.com/pub/ailogblo...

News update: The AI revolution is still not happening. The California State University System and OpenAI issued a press release. That’s all that happened on Monday. Nothing more. What exactly is an AI-Empowered University System? open.substack.com/pub/ailogblo...

We all needed a distraction from the kleptocratic fracking going on in DC, but DeepSeek just isn’t that good. I don’t mean how the model performs on benchmarks. I mean as a plot twist in the remake of Silicon Valley that was playing on all the channels this week. open.substack.com/pub/ailogblo...

Rob Nelson, who currently holds the title belt for best commenter on my Substack, has some interesting insights on using AI in his teaching here. The point about the teacher not always being best feedback provider is worth ruminating on.

One reason I used an LLM to teach last fall was that I wanted to offer my students an alternative to ChatGPT. Another was my curiosity about what educational value, if any, an LLM tool might add to my active in-class learning approach to teaching. open.substack.com/pub/ailogblo...

I wanted to experiment with LLM tools in my history class, but I don't have the technical chops to roll my own LLM wrapper and I didn't want to use a commercial product. What is an LLM doing in my classroom? Part 2: Experiments open.substack.com/pub/ailogblo...

I gave my first talk of 2025 on Friday. Since I didn’t teach in 2023, this was an opportunity (and a deadline!) to organize ideas about what changed in my teaching practice post-ChatGPT. What is an LLM doing in my classroom? open.substack.com/pub/ailogblo...

Once again, @ddmeyer.bsky.social is doing work in his spare time that tech journalists should have done. The Truth About 2 Hour Learning & Unbound Academy a/k/a The School “Replacing Teachers with AI” open.substack.com/pub/danmeyer...

Do our grading practices harm students? Are schools and colleges forms of artificial intelligence? Can we change the purpose of our big algorithmic educational systems from sorting humans to helping them grow? Dan Davies and @josheyler.bsky.social help answer. open.substack.com/pub/ailogblo...