OLWeekly

Stop Trying to Make AGi Happen, It’s Not Going to Happen!
Matt Crosslin, EduGeek Journal, 2025/03/21

Matt Crosslin responds to George Siemens. He first crtiticizes one of the core points: "a system made up of many individuals has more intellectual capacity than individuals? Well… duh. Hasn't it been that way since the beginning of human intellect? Adding more people gives you more capacity." I don't think this captures the nuance between simply stacking capacity (ie, adding volume) as opposed to combining capacity (ie, adding complexity). And when Siemens writes, "Personal knowledge is comprised of a network, which feeds into organizations and institutions, which in turn feed back into the network, and then continue to provide learning to individual. This cycle of knowledge development (personal to network to organization) allows learners to remain current in their field through the connections they have formed," Crosslin responds, "None of this is artificial." But I don't think we've seen a naturally occurring organization or institution through all of history. But Crosslin's main point is in the title: "I am pretty skeptical that AGi will ever happen – but people keep trying to see it where it doesn't exist." I'd defer to his deep knowledge here, but if he makes basic errors, I think it might just be an opinion based on an intuition, and nothing more.

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

I've seen amazing mixed reality experiences, but a new update will unlock the next generation
Nicholas Sutrich, Android Central, 2025/03/21

The gist: "this is one of the most convincing uses of mixed reality to date. Most people don't have access to an empty airplane cockpit in their garage, but being able to turn your favorite chair into a convincing airplane cockpit - complete with its texture accurately positioned inside the virtual cockpit - would make an incredibly convincing game of Flight Simulator." Maybe. But it's still a device that will only be used occasionally, and never in combination with other things (like, you won't be watching a movie while playing mixed reality games, though I do that a lot with regular video games).

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

Equation Editor for online mathematics - create, integrate and download
CodeCogs, Zyba Ltd, 2025/03/21

This is pretty interesting, a "LaTeX equation editor that creates lightweight graphical equations (gif, png, swf, pdf, emf) and produces code to quickly and directly embedding equations into HTML websites, forums or blogs." Via Alan Levine.

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

What do we mean when we talk about ‘openness’ in (generative) AI?
Doug Belshaw, Open Thinkering, 2025/03/21

The main value of this post is its introduction of the Open Weight Definition, which is more of a technical definition of openness than a cultural one (to paraphrase Belshaw's own phrasing). I think the article could use a third section, and could talk about open data - is it important? (I get the sense Belsaw thinks not) and what role does it play into open AI?

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

A Theory of Argumentative Success
2025/03/20

What is it for an argument to be successful? John Keller writes that as "argument evaluation is central to the philosophical enterprise" it's important to be clear about what that means. To this end he offers a taxonomy of theories of success and defends an 'knowledge account of epistemic success,' that is, "the concept of an argument that 'works'... is that arguments can establish their conclusions for certain people without establishing them for others, and that an argument establishes its conclusion for someone iff that person knows it to be sound and non-fallacious." The argument, in other words, must be truth-preserving, for truths as known by a given individual. That sounds find, but if we accept it, then we need to go back to the original assertion and ask whether that's what philosophy should be doing? And that is not so nearly clear to me.

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

23 good reasons Archives - unitwin-unoe
unitwin-unoe, 2025/03/20

I criticized Rob Farrow earlier this week without really understanding that his was just one of 23 articles in a series called "23 good reasons for Open Education" (in my defense, you really have to be tuned in to see that, since the articles lack any series introduction, and even the title of the series is abbreviated in the header). The reasons include things like "build a sustainable education system" and "incentivize original thinking" and "recognize teachers' work". While I think the list might be a list of good things about open education, I think that offering some of these as 'reasons' to support it is a bit of a stretch. If you're looking to "recognize teachers' work", for example, there are many better ways to do it. Like paying them more money, for example.

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

11 Thoughts About the Massive Layoffs at the U.S. Department of Education
Michael J. Petrilli, Education Next, 2025/03/20

My feeds have been filled with concerns about cuts at the U.S. Department of Education, understandably, but it's important to highlight that in many circles the cuts are greeted with indifference and even enthusiasm. Michael J. Petrilli (Fordham Institute president and visiting fellow at Stanford University's Hoover Institution) writes, for example, "I don't anticipate these staff cuts will make much of a difference in the real world of schools and classrooms." Sure, he's paid to write stuff like this. But large numbers of Americans are buying what he's selling, and it's important not just to stand against education cuts but to understand why the cuts are supported by so many people.

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

7 customer experience trends in 2024
Alexandria Iacoviello, IBM Blog, 2025/03/20

This is an article that gets the overall trends right, but which also gets each one wrong in some telling ways. You'll see what I mean. Trend 1 speak to "sustainability" as a business advantage. It's about belief "that the brand cares about its impact on people and the planet." But nobody believes brands 'care'. Or Trend 4, "prioritizing customer accessibility to information", instead of, you know, providing information. Or Trend 5, "personalized experiences using data and generative AI," instead of personalization by enabling personal choice. Or Trend 7, "ensuring data privacy for customers," instead of ensuring customers own their own data. Right issues, wrong emphasis.

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

Taking natural-looking motion to yet another level
Doug Belshaw, Thought Shrapnel, 2025/03/20

The company that brought you the robot dog is now demonstrating the robot human. The demo model, from Boston Dynamics, is more athletic than I am. As Doug Belshaw says, "The usual response to this kind of thing is amusement tinged with fear." Yes, it could be a lethal weapon. But just add it to the pile of lethal weapons that already exist. As usual, the real danger here is the human operating the system. Meanwhile, "imagine the robots in these videos doing the jobs that currently require humans doing things that might endanger their health, such as rescuing people from burning buildings, inspecting nuclear reactors, and even doing very repetitive tasks under time pressure in warehouses." Oh, and as robot teachers.

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

British Council accused of forcing gig economy teachers into 'feeding frenzy' for work
Tom Wall, The Guardian, 2025/03/19

Instead of trying to define change in ways that are 'comfortable' for educators, we should perhaps be thinking about how to prevent something like this becoming the norm: "The British Council has been accused of exploiting hundreds of agency teachers on zero-hour contracts forced to compete for lessons in a 'feeding frenzy' every week... up to 350 teachers based in the UK, US, Canada, Australia and India have to race each other to book fluctuating numbers of classes released every week." It's not hard to imagine a '2025 vision' where public education at all levels is replaced with an 'Uber of learning' for the majority (and lavish personal education for the privileged few). See also: Current Affairs.

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

AI and the Calculator Effect
Steve Hargadon, 2025/03/19

"Lately," writes Steve Hargadon, "I've noticed something unsettling: younger people I encounter - cashiers, students, and others - struggle to do relatively simple math in their heads." Naturally, the argument is extended to AI: "Today we're seeing a similar loss in intellectual capacity with the use of AI, specifically large language models (LLMs). A spate of recent articles has documented how the use of LLMs can have the effect of reducing the writing and thinking capacity of both students and adult workers." I don't think we've seen this reduction in the writing and thinking capacity, certainly not as a result of the use of AI; it simply hasn't been around long enough to have had that effect. The articles are projecting a loss of capacity, not reporting on one. But here's the thing. In my 20s, I worked in food services and retail, and at that time, I noticed people struggling to do simple math in their heads. That was in the 1970s, before calculators were available. I also notice people don't solve cryptograms in their heads, like I do. Because of writing, people don't think in their heads. Not until they really need to (or want to). So don't blame it on tech.

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

Should higher education be thinking in terms of evolution or transformation?
Debbie McVitty, WonkHe, 2025/03/19

I like the way this article argues that there are different types of change, with different actors implicated, and at different speeds, though I think the discussion here in incomplete and serves to comfort educators perhaps more than they should be comforted. After all, while there's no doubt many types of evolutionary change happening, and while these changes are gentle and calming, the way things are going it seems likely that they will be overwhelmed by much larger changes not anticipated in this article. The list of transformational changes - which include federation, harmonization, and acquisition - is missing a crucial possibility: annihilation. None of the 'transformational' changes anticipate, for example, a 50% or greater budget cut. But these days, especially worldwide, that's certainly a possibility (not one, mind you, that I endorse; but preparing for a pandemic doesn't mean you're cheering for the virus).

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

Developing a custom hugo theme with claude
D'Arcy Norman, 2025/03/19

I mean, I can't let this post go uncommented, presented as it is in a style that couldn't be more opposite to my own tastes, a sort of mustard Courier font against a black background, resembling the CRT monitors of yore. Especially when the topic of the post was how that design was achieved for Hugo - a static web development tool I don't use - using Claude. To be sure, the mustard is better than the lime green originally selected, and it is kind of easy on the eyes to read, but there's no way to present images in it without a garish clash of format. And the spacing and text size are actually pleasing to the eye, but only because Courier is not really an easy-to-read font. Opinions on design are strictly my own of course :) and I like the way this is all explained. (p.s. D'Arcy, the map command in the meta author element isn't executed, it's just printed in plain text; see your source).

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

OpenAI Wants to Ban DeepSeek
Michael Spencer, AI Supremacy, 2025/03/19

According to this article, " BigTech wants to use U.S. anti-China rhetoric and U.S's push for AI Supremacy to veto Open-source models that might threaten its own global dominance and cash-cows." Martin Dougiamas opines in Mastodon, "Michael Spencer here encapsulated some of the disgust I feel about #OpenAI distinctly anti-Open monopolistic intentions (which is part of a much larger problem of greed in US corporate culture)." My take is that AI has been for decades a global endeavour, that it should be an open endeavour, and that nobody really 'owns' the basic principles researchers have worked out. And these days, I feel no more sanguine about U.S.-based AI than Chinese AI.

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

Favour equity
Robert Farrow, EUniWell Open Education, 2025/03/19

Steve Boots ran an interesting video today on why the right dominates online media. One good reason, of course, is that it is much more difficult to find financial and (especially) corporate support for left wing media, since the left promotes social values in addition to, or instead of, economic values. But another reason is equally relevant: writers on the left are much more likely to criticize each other for (say) not being sufficiently left, or left in just the right way. Much the same could be said of proponents of open learning. Now I don't mean to make an example of Rob Farrow here; this is just the next article I happened to encounter. The premise here is that the reason to support openness is to support "the struggle for equity in education". It's important, in this article, to support openness in the right way: "While open education offers a promising pathway toward greater equity, it too must reckon with its own limitations and biases, leveraging openness as a critical lens on existing and historical practice." I think we can be looser in our requirements. If someone supports openness, I think that's great, and don't feel the need to put them through a values test.

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

Complete L&D Content Curation Toolkit
Mike Taylor, Mike Taylor, 2025/03/18

Any time I see a formula like this (and we just saw one from Next Gen News) I think of the old aggregate - remix - repurpose - feed forward methodology that I've often described as characteristic of connectivism. Not that this characterization was anything original either. It's what anyone - or any thing - in an information ecosystem does. This curation model from Mike Taylor is a little more focused on intention than most. Curation has to have an objective, a reason for being done. He also adds a bit of a toolkit for the curation process that is a bit more fill-in-the-box focused than I'd like. It's the difference between giving a piece of content a 1-5 rating, and writing something free-form (like I do) about the content.

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

Wired’s un-paywalling of stories built on public data is a reminder of its role in the information ecosystem
Joshua Benton, Nieman Lab, 2025/03/18

Wired magazine announces that "it's going to stop paywalling articles that are primarily based on public records obtained through the Freedom of Information Act." That's the right stance to take. Of course it's a lot easier to take that stance in an environment where the National Center for Education Statistics, which until January 20 had about 100 employees, now has only three. See also: Freedom of the Press Foundation.

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

EduChat: A Constrained LearnLM Tutor
Jim Salsman, Replit, 2025/03/18

Jim Salsman writes, "I spent some time making a Streamlit app to demonstrate how to use them with system prompts providing instructional support that has been weakened in that latest LearnLM model, but can be substantially restored in it and their other models, too." This link is to that deployment. It's hosted on Replit and connects to a free Google account and Google's LearnLM, so isn't production-ready, but it's a great illustration of the concept. I tried it out - it adopts some well-known instructional strategies, such as breaking down the topic into more manageable chunks and asking the learnerr to respond with their own thoughts. Here's the source code. Image: Dataconomy.

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

Docs - Open source alternative to Notion or Outline
Hacker News, 2025/03/18

This post links to the code on GitHub for Docs, an open source replacement for Google Docs developed by the French and German governments. Here's the Docs website. It's available online but you need to belong to a French professional organization to login using the French ProConnect service (Docs uses OpenID Connect (OIDC), so if you already have an OIDC provider, obtain the necessary information to use it.).

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

Next Gen News
FT Strategies, Knight Lab, 2025/03/18

The research basis isn't strong (45 interviews with youth in Nigeria, India and the U.S. plus comments from a 19-member advisory board) but the result is interesting and I think relevant to educators as well. There's an HTML version you can click and scroll through or a single file (84 page PDF) you can download. The ideal news experience comes from a trusted source, is personally relevant and actionable, and convenient to access. It offers a five-modes framework - sift, substantiate, study, socialise and sensemake - that won't feel out of place to readers here. There's also a Next Gen News toolkit (presentatioin view, Google doc) facilitators can use to lead a session based on the report.

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

Tending Tomorrow's Soil: Investing In Learning Ecosystems
Gregg Behr, Getting Smart, 2025/03/17

"Real-life learning ecosystems – networks that include classrooms, living rooms, libraries, museums, and more – may be the most promising approach for preparing learners for tomorrow," writes Gregg Behr. What's needed is "a collaborative ecosystem that sparks engaged, motivated learners by providing the tools, resources, and relationships that every young person deserves." I would say 'cooperative' rather than collaborative, and I would focus much more on public support for such an ecosystem (because private support would invariably commercialize it). Now for the other shoe to drop: "These ecosystems are not replacements for schools," writes Behr. But inevitably, they will be, as public education systems struggle with declining resources and increasingly complex societies.

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

The new consumer decision-making process
Think With Google, 2025/03/17

As much as educational institutions may not want to talk about it, they to a degree depend on sales, in the form of student registrations. This article reconsiders the traditional 'sales funnel' which runs from awareness to consideration to action (it's called a funnel because the number of people gets narrower and narrower through each state). This article proposes a '4s' version of the funnel: streaming, which refers to the new 'always on' internet people have; scrolling, which refers to their browsing behaviour; search, which refers to their consideration of alternatives; and shopping, which refers to the actual purchase.

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

Nobody is Coming to Save Us, But...
Alex Usher, HESA, 2025/03/17

"The age of American Science is over, and it's not coming back any time soon," writes Alex Usher. "The opportunity exists, therefore, for ambitious universities to scoop up a fair bit of top new talent." He's speaking to a Canadian context, but of course the opportunity exists worldwide. And in Canada, since governments won't find new researchers, he suggests, universities should consider recruiting foundations to fill in the gap. The problem with this is that foundations have an agenda, and in a lot of cases that agenda lines up with the sort of thinking that led the U.S. into the position it's in now. There's a reason why we want public funds to support public education; it keeps the country from going off the rails.

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

Manus
2025/03/17

George Siemens links to a couple new language models from China in his latest newsletter. First is Manus, "a general AI agent that turns your thoughts into actions" by deploying AI agents to do tasks for you. "Here's an example of a 'big tech stock performance' I requested," writes Siemens (noting it's hard to track how accurate the output is). Alsoi from China, Baidu launched its own "4.5-worthy LLM", Ernie. "Biggest difference between Ernie 4.5 and GPT-4.5? Ernie is 1% of the cost."

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