OLWeekly

Is language the same as intelligence? The AI industry desperately needs it to be
Benjamin Riley, The Verge, 2025/11/27

"According to current neuroscience, human thinking is largely independent of human language." I believe that this is true but I don't see it as an argument against AI (though it does show that large language models will not in the long run give us the AI we want). "Three scientists published a commentary in the journal Nature titled, with admirable clarity, "Language is primarily a tool for communication rather than thought."

Web: [Direct Link] [This Post][Share]

Who's Grading You on Coursera? The Shift from Human Peers to AI
Pat Bowden, Class Central, 2025/11/27

Coursera has two problems. The first is that its peer assessment system is broken. The second is that it needs to continue to scale and attract investment. Both of these are leading it to turn to AI. "People are eager to invest in the parts of a business that scale. Not everyone wants to invest in the humans to maintain that scale. Peer review is slow and can be messy and expensive to do well." The core question is this: does this end the slow decline of Coursera's credibility, or does it accelerate it?

Web: [Direct Link] [This Post][Share]

Alchemy
Josh Collinsworth, 2025/11/27

I think these two statements say very different things: first, that "Art is valuable precisely because it is not easy to create," and second, that "We care about art because it';s a form of connection to other humans." Now, true, sometimes it's hard to connect with other humans, but sometimes it's all too easy (at which point, we begin to value insulation from other humans). I really don't care how hard it was to produce something, at least, not intrinsically. I do value what doing something hard tells me about the person. And none of this tells me how much I should pay for a thing, because all of this is cheapened if it just becomes a form of transaction.

Web: [Direct Link] [This Post][Share]

How citizen archivists in South Asia confront the online marginalization of oral cultures and languages
Amrit Sufi, Diff, 2025/11/27

One day I will wrestle with the idea of how there can be multiple knowledge communities, but for now, I wish only to emphasize the need to preserve them. "Documentation of these narratives by citizen archivists who are native speakers of the languages ensures that these are not excluded from the mainstream knowledge forms online, thus supporting epistemic and social justice." Related: Pluriversality of Knowledge in the Age of AI. "What will happen to the knowledge of those who cannot speak or cannot live, those who are unable to own or govern the record of their memories and experiences?"

Web: [Direct Link] [This Post][Share]

From detection to development: how universities are ethically embedding AI-for-learning
Mike Larsen, HEPI, 2025/11/27

"I believe the 'Police and Punish' model must now be replaced," writes Mike Larsen. "A reliance upon detection was perhaps once a necessary evil but it has never aligned with the fundamental values of higher education. The assumption that policing student behaviour is the only way to safeguard standards no longer applies." So what to do instead? "An emerging policy framework for consideration and research is 'support and validate' which pairs timely, evidence-based academic support with student self-validation of authorship and learning."

Web: [Direct Link] [This Post][Share]

When grades stop meaning anything
Kelsey Piper, The Argument, 2025/11/26

Fill in the blank:  7 + 2 = [_] + 6. If you're like me, you instinctively said '1' and would have been among the 25 percent of students who got the question wrong in the University of California San Diego's (UCSD) remedial math class. Now, of course, I know (and so do you) that the correct answer is '3' and there's no chance that I would really be wrong about this. So what happened? My take is that there's something about the way the question is posed that leads to a quick wrong answer. I think a lot of math - and probably a lot of math testing - is like this (though I have no real evidence for that statement). It's like when people say "No one can do fractions." I think they can, they just aren't given the tools (simple things, like asking "what is 1/2 of 1/4" instead of "what is 1/2 times 1/4" (in both cases, it's 1/8)). I struggled with calculus, for no good reason other than that - it was presented to me as a bunch of stuff I had to remember, and not a bunch of stuff I should understand. Kelsey Piper says "Cargo cult equity needs to die." But I think we need to think more seriously about what equity actually means.

Web: [Direct Link] [This Post][Share]

Introducing Nested Learning: A new ML paradigm for continual learning
Ali Behrouz, Vahab Mirrokni, Google Research, 2025/11/26

This article is a summary of the full paper (16 page PDF) describing what may be a significant advance in AI. It's called 'Nested Learning' and is described as follows: "Nested Learning treats a single ML model not as one continuous process, but as a system of interconnected, multi-level learning problems that are optimized simultaneously. We argue that the model's architecture and the rules used to train it (i.e., the optimization algorithm) are fundamentally the same concepts; they are just different 'levels' of optimization, each with its own internal flow of information ('context flow') and update rate." Though they deal with different subject matter, all of these layers are based on the same principles of associative memory. I admit I've been a bit slow on the uptake here (the paper is a couple of weeks old) but I've been swayed by reaction to it (with one writer calling it "Attention Is All You Need (V2)").

Web: [Direct Link] [This Post][Share]

Understanding Learning Strategy Use Through the Lens of Habit
Ann-Kathrin Krause, Jasmin Breitwieser, Garvin Brod, Educational Psychology Review, 2025/11/26

Maybe it's just me, but I always find a deep disconnect when I read studies like this talking about effective learning. Here, for example, we read, "students frequently rely on ineffective learning strategies instead of those that promote long-term retention." This is meant as a criticism of (so-called) 'self-regulated learning' (I prefer the term 'self-managed', but I digress) which asserts, essentially, that self-managed learners develop bad habits. Well, OK, I can see this. But what are the strategies being considered? The 'good' student (leading to better test results) are "spaced study sessions over time, tested herself, and elaborated on study material" while the bad student "relied on rereading and underlining in order to process material." Now, I don't know anyone who thought underlining would lead to remembering. It was detect patterns. My own experience is that, if you want to remember something, make it easier to remember by identifying the structure (not the same as a concept map, though I did that a lot; it's more of a memory palace effect, but using the work itself as the palace, a la Keith Spicer's Winging It). That's what launched me from a pretty good test taker to an expert test taker. Now this became a habit with me, and has served me exceptionally well over 45 years. So I feel quite disconnected with the article (30 page PDF). Via Robert Gibson.

Web: [Direct Link] [This Post][Share]

How Big Tech is quietly colonizing education
John Moravec, Education Futures, 2025/11/26

According to John Moravec, frameworks for the use of AI in education normalize the use of AI in education, and at the same time, they distance teachers and administrators from the decisions being made about how and why it is so used. "In a century shaped by powerful digital systems, education cannot limit itself to procedures for safe use. It must reclaim its role as a steward of human judgment and collective purpose. That work begins when educators refuse to let AI define the terms of its integration and instead place education's values at the center of the conversation." That's fine, but, whose value is that, exactly?

Web: [Direct Link] [This Post][Share]

Introducing Nano Banana Pro
The Keyword, 2025/11/26

People have been expressing a lot of enthusiasm for the new Nano Banana image generator AI from Google. Here's Mike Caulfield, for example, with a nifty infographic from his 52 video walkthroughs of Critical Thinking with AI Mode. Here's a bunch more. You can make professional infographics with it. Now I've seen a bunch of these generated infographics and the technology is really impressive. But... what exactly is it representing? I thought I'd try with my 1,000 page manuscript on ethics and AI. I first loaded it into Gemini, thinking it was Nano Banana, and it created an interesting 'BANANA' acronym to represent the contents of the document, and does a pretty good job. Then I produced the image (view it here) and while it represents what I probably should have written to fit squarely into the mainstream (and the path to riches and fame) it completely misrepresents what I actually wrote. So there's more happening in Nano Banana than mere analysis of the documents being depicted as infographs. Much more.

Web: [Direct Link] [This Post][Share]

Pastoral Landscapes Brim with Patterns in Luminous Paintings by David Brian Smith
Kate Mothes, Colossal, 2025/11/25

Remember 'dogception'? That was (as illustrated in this post) what resulted when an AI over matched for patterns resembling parts of dogs, thus reinterpreting any impage so it looked like it was filled up with dogs. I get a similar effect when I look at my bedroom ceiling - I 'see' dogs in the sames and patterns of the knots in the pine boards. This articxle describes David Brian Smith's similar sort of effect in pastoral landscapes - not just dogs, but all matter of things. "Within the sky, fields, rivers, and forests, hundreds of little hatch marks, flowers, starbursts, and other thematic motifs dance across the surface." I don't see the world that way (I wish I did) but I can easily understand how someone would, and how it might alter their sense of what things actually are.

Web: [Direct Link] [This Post][Share]

Replicability and the humanities: the problem with universal measures of research quality
Chloe Patton, Research Evaluation, 2025/11/25

This is an excellent article that speaks to me specifically as someone based in the humanities who does work in the field of education. It argues that replication - using new data to solve the same problem - cannot be applied to the humanities because there is not an independent reality against which the studies can be compared. Rather, "The central ontological assumption at the heart of most research in the humanities is that reality is socially mediated through language." That does not mean that 'anything goes' in humanities research (though it does argue against a formal research methodology). Evaluation of argumentation in the humanities employs "all manner of critical tools employed to strip the flesh from ideas and expose weaknesses in interpretations." Now in my own work I take the concept a bit further: I don't think knowledge need be 'socially mediated', and I think there are many other media over and above language. What then can I usefully say about education? Well, quite a lot, I think. Not replications of perceptions as expressed through Likert scales, nor assessments of 'what works' in education, but rather, ways of looking at the means, motivations and assumptions that lie in the implementations - or rejections - of educational technology. Image: UQueensland.

Web: [Direct Link] [This Post][Share]

Exploring Joplin Note-Taking Tool
Miguel Guhlin, Another Think Coming, 2025/11/25

Miguel Guhlin reviews Joplin Notes, a new online note taking application that can be run locally or in the cloud. It's one of a class of tools designed to organize your writing when you write a lot of content. The main right-up-front feature is easy conversion from markdown to HTML. 

Web: [Direct Link] [This Post][Share]

Anatomy of an AI agent knowledge base
Bill Doerrfeld, Infoworld, 2025/11/25

This is a useful article in that it fulfills its stated purpose: it gives the reader a good description of what a knowledge base looks like and how each element is used to make the AI that uses it more responsive and accurate. But what struck me as I read it is that it offers a good analogy to a human's knowledge base - there is the set of guides, conventions and rules that "mirrors what you';d find in a senior employee's mental toolkit." There's the data in a database. There are policy and procedure manuals. And then there is the semi-structured knowledge equivalent to a knowledge wiki or even a personal library. Each person's knowledge base is unique, and each has their own 'data moat' - the distinctive knowledge that gives them value in the workplace. Via Miguel Guhlin.

Web: [Direct Link] [This Post][Share]

A New Era Begins
Doc Searls, Doc Searls Weblog, 2025/11/24

As described here by Doc Searls, "MyTerms describes how the sites and services of the world agree to your terms, rather than the other way around." To start the process, a set of five simple sets of terms are made available (a model popularized by Creative Commons). You can choose between 'service delivery only' (aka SD-BASE), 'data portability', 'data use for AI training', 'data for good', and 'data for intentcasting'. Of course, with websites like mine, I don't collect any data, there's no cookies, no accounts, no surveillance and definitely no adtech. If you want me to send you email, you tell me your email address. That's it. You don't want email any more, I forget your email address. The point of 'MyTerms' (like Creative Commons) is to facilitate commercial use, under the heading of 'free'. My version of 'free' is 'free from all this'.

Web: [Direct Link] [This Post][Share]

The Degree That Never Ends
Carlo Iacono, Hybrid Horizons, 2025/11/24

This is an article that could stand being half its current length but which makes a useful point. Universities seeking relevance are often advised to "transform the university from a four-year credential factory into a lifelong learning partner." Sounds great, but at a certain point learning, properly so-called, ceases to be an opportunity and begins to be an obligation. "When learning never ends, that space collapses. Every book becomes a potential credential. Every skill a line item on a digital passport. Every moment of growth an opportunity to document, to verify, to add." Or in summary, "Lifelong learning promises empowerment but delivers perpetual dependency... Your expertise is not embodied but rented. Your wisdom is not developed but subscribed to. Extraction dressed as service."

Web: [Direct Link] [This Post][Share]

New Token-Oriented Object Notation (TOON) Hopes to Cut LLM Costs by Reducing Token Consumption
Bruno Couriol, InfoQ, 2025/11/24

As this short article reports, "the recently released Token-Oriented Object Notation (TOON) aims to be a schema-aware alternative to JSON that significantly reduces token consumption at a similar level of accuracy." Basically, what it does is to remove all the extra quotation marks and punctuation in JSON and well as removing the duplication of field names in individual records. It points (in my view) to the need for a better AI pricing model.

Web: [Direct Link] [This Post][Share]

The existential struggle between being a 'we' and an 'us'
Tris Hedges, Psyche, 2025/11/24

In Black Like Me John Howard Griffin wrote about 'the hate stare'. It is what Sarfaz Manzoor calls one of the "routine torments of discrimination." It is an example what Jean-Paul Sartre called 'the look', an objectifying glance that classifies you without perceiving you. It's not necessarily a look of hate; it can be subject to any number of emotions, but it always feels distancing. This article captures the same concept from a different direction. You and I are going for a walk; we are going for a walk, and our collective experiences are felt through our embodied subjectivity. 'We' is described through "referring to the plurality as the subject (of an action, belief, judgment, emotion, perception, etc)." But when we are seen by a third party, that person sees us, and being seen this way is a very different kind of collective experience, what W.E.B. DuBois calls "this sense of always looking at one's self through the eyes of others, of measuring one's soul by the tape of a world that looks on in amused contempt and pity." How many of our experiences in education are of 'we' and how many are of 'us'? Image: from my film Bogota.

Web: [Direct Link] [This Post][Share]

Job for 2027: Senior Director of Million-Dollar Regexes
Tim O'Brien, O'Reilly, 2025/11/24

This short post demonstrates the difference between having AI do something for you, and asking AI how to do that thing. Imagine wanting to remove all the social security numbers from a million documents a day. Using ChatGPT 5, that could cost $6 million a year, accoring to this article. But if you ask ChatGPT how to do it, you be supplied with a regular expression (regex) that can run in any old computer code for a fraction of the cost. "Starting now, you'll be able to make a career out of un-LLM-ifying applications," comments Tim O'Brien. 

Web: [Direct Link] [This Post][Share]