
That is to say, AI detectors don't belong in education, period. If you're a professional, writing for a public audience, and vast quantities of your writing gets flagged as almost entirely LLM output, you're done

Literary critic in a library. Egoism and the avant-garde. Anarchistique. Views all my own.

That is to say, AI detectors don't belong in education, period. If you're a professional, writing for a public audience, and vast quantities of your writing gets flagged as almost entirely LLM output, you're done

pretty soon tech bros are going to claim they invented hubris

This is how machine learning (ML) should be applied: Have a reliable system X working in some area Y. Distill a faster simpler ML model M from X under conditions C. Apply M to Y under strict constraints that conditions C are satisfied. But most folks want to use M outside C.
It also always relies on a gold standard model to imitate. So it's incapable of solving novel problems. You have to build the thing that solves the problem first, and then you can use ML to imitate that solution.

Large language models (LLM) are as smart as smartphones are smart with the caveat that smartphones are not actually smart. Don't confuse mimicry with intelligence. LLMs + harnesses can be useful but intelligent, they are not.
This is a problem that almost no one talks about, and I'm struggling to explain it without people jumping to the wrong conclusions. Clearly, LLMs are getting way smarter at a startling pace. So there's this group, AI doomers in particular, who believe intelligence is one-dimensional. They're convinced that as LLMs keep getting smarter, we must be approaching AGI, and we're all doomed. Then there's another group, CritRats in particular, who seem to believe intelligence is zero-dimensional, or binary. You either have it or you don't. So none of this progress tells us anything meaningful about general intelligence. The real story is that LLMs were a huge breakthrough. Yes, they even represent a significant step, or at least a stepping stone, toward AGI. If nothing else they taught us the universality of human language. But then a tremendous amount of effort shifted toward making LLMs useful through increasingly sophisticated harnesses. We've certainly continued training better models, and that's a very good thing. But much of the impressive progress has come from training them on specific, verifiable tasks like mathematical proofs and coding, where we have known correct answers to train against. This has been huge in terms of making LLMs genuinely useful. But it does not make them 'more general' or the like. Then there are reasoning models, which essentially tell LLMs to stop and think things through before answering. Again, a huge improvement in practical usefulness, even if it doesn't necessarily represent a fundamental improvement in the underlying model. But LLMs are not true general intelligences. And I don't think we're on a trajectory toward making them so. Likely there is just no way to do so because they are the wrong kind of intelligence from the outset. That doesn't mean they aren't incredibly important. You don't need true general intelligence to disrupt the world. In fact, there may be considerable advantages to having powerful intelligences that aren't general intelligences: they're essentially robots that, for the most part, just do our bidding. And perhaps it's even better that way for many tasks. Here's what I think people are missing: We can keep improving LLMs indefinitely, discovering one incredible new application after another, with one breakthrough after another, without ever actually attempting to create true general intelligence. There is an enormous amount of unexplored territory in this kind of intelligence. We should explore it! Even if we eventually invent true general intelligence, there will still be enormous value in continuing to improve LLMs as tools. We'll probably be improving them forever. This would be possible even if (as I'm guessing) they aren't a path to human level intelligence. Intelligence isn't one-dimensional. There is plenty of room for extraordinary progress even though we're not yet on a trajectory toward inventing general intelligence.

AI hallucinations are built in and cannot be overcome. The "revolution" has already hit a plateau. There are only incremental tweaks from here on.
If hallucinations are a built-in insurmountable part of the technology, AI will never be reliable and there won't be the AI revolution some people predict. And there's no evidence hallucination can be overcome

We have broadly illiterate parents raising children who don't read. Broadly illiterate teachers educating those children. Broadly illiterate education academics training the teachers. Broadly illiterate executives pushing AI into the education system. It ain't the kids, hey

We are sleepwalking into a disaster. Let me share a real story from my wife, a professor at UGent. She mentors a new generation of PhD candidates. Some are producing excellent results, even improving state-of-the-art algorithms, yet struggle to explain what the original papers actually do, let alone what their own contributions add. After days of carefully reviewing a paper, she sends back comments and asks for corrections. Thirty minutes later, she receives a response that is nothing more than AI-generated text. I suspect this is not an isolated case but a symptom spreading throughout academia: a new generation of “researchers” who can produce results without understanding them, improve algorithms without grasping them, and write papers without truly knowing what they are writing about. We risk creating a generation that knows how to produce research but not how to do research.
数学家 Michael Goldman 昨天发了篇文章,标题就很狂:《我们还需要数学期刊吗?》他自己的答案是:不需要。 最狠的一句:他现在更信任 LLM 的审稿,胜过人类审稿——包括他自己审的。 他还算了笔账:法国公共科研一年花约 7800 万欧元订学术期刊,所有学科加起来,一半进了 Elsevier 的口袋。这笔钱,够给全法国十几二十万研究者人手一份 LLM 基础订阅,还能剩 2000 万。 那期刊现在到底还剩什么用?传播有 arXiv;把关他更信 AI;筛选靠同行讨论,正式见刊要等一年多。 拆到最后,期刊只剩一个用途:招聘和晋升的时候数论文。😂 所以一个一年 7800 万欧元的系统,最后的产出却是一个 KPI。

Whoops! Turns out that the rector of @PUT_Poznan, Teofil Jesionowski actually has another retraction!👇This #papermill #fraud is closely linked to the biochar cartel and their network of frauds! These people are destroying science and academia! doi.org/10.1016/j.envres.202…
Major news! the rector of the @PUT_Poznan, the #fraud Teofil Jesionowski got his first retraction! Notice also Jakub Zdarta, an amateur fraud and the professional frauds Hafiz M.N. Iqbal & Muhammad Bilal! Frauds always rise to the top. doi.org/10.1016/j.colsurfb.2…

This includes a 72 year old Jewish person who was arrested and charged.
QLD police arrested 6 people in Bris on Sunday who were protesting recent laws banning words “from the river to the sea”. If you haven’t heard about these arrests, it's likely because we can’t find any media reports on it. Police haven't included it in its daily press releases

i love pro-AI’ers who refute arguments re: labour and environmental impacts by being like “well we already allowed so much automation so why not AI” or “we eat the products of industrial agriculture, so why not data centres” everything is shitty so we should pile more shit on top

If you think the AI data centre build out in Australia is slowing down, Nvidia is signalling exactly the opposite


I helped to organise a symposium on the political aeconomy of Artistic Freedom in Australia in June. Here is a collective statement from participants kindly published by @OverlandJournal: overland.org.au/2026/11/who-…

See this from Virginia Woolf.
Virginia Woolf’s essay on how to read books and different writers is fantastic. She does so well in explaining that you need to give in to the author’s style and intention as if entering into a new world. yalereview.org/article/virgi…

I’ve secretly always wanted to grade student writing by a single criterion: “Was I bored out of my fucking skull?” I feel like gen AI is giving me permission to finally implement this.

The AI mindset rests on a fatal flaw: the belief that all problems can be solved by more information. This is Dataism. It can be disproved by the problem of acute loneliness. Mapping the neurological & social data of loneliness does not translate into solutions or remedies. In fact our reduction of humans to data sets is increasing our social atomisation & loneliness.