Stephen Downes

"To teach well," writes David Webster, "is not simply to transmit knowledge or to facilitate the production of correct answers. It is to model a way of being in relation to knowledge: one that is attentive, discriminating, and alive to the possibility that understanding can deepen." This has implications, he writes, that tend toward a system of education based on presence, that is relational, iterative and slow. Fair enough, and I'm all for taking time with nuance and precision (though I think attitudes like A. J. Liebling's "kind of delighted voracity... to seek out, taste, compare, and pursue the best that can be found" is more pretentious than anything). The problem with Webster's argument is that while learning slowly is fine, teaching slowly is inefficient and expensive. What we need are ways (and means, and motivations) to be able and willing to learn slowly on our own for otherwise most of us will not have the opportunity to learn at all.

Apostolos Koutropoulos responds to an article in TD magazine (both link and archive link point to a paywall, so don't bother). He writes, "There are two threads here. In one thread, I feel like the author is trying to get a bit Vygotskian here and treating the LLM like a more knowledgeable other, which they are not... The other thread is that the LLM becomes a mirror for your thoughts. Cool, I guess, but do we really need a modern-day ELIZA." But the main complaint is that it seems to treat the failings of the LLM as the fault of the user. Now granted, LLMs are not perfect. That doesn't mean they should be written off completely. They do some things very well. And to that extent, the skill of the user does come into play, if only to use them for tasks where they excel, and not where they don't.

This article introduces the 8-SLDF, "a framework that serves as a comprehensive guide to the systematic development of formal curricula in tertiary education." The primary value is it's placement of the framework in a series of entities ranging from 'philosophy' and 'principles' through to 'models' and 'patterns'. It's all very structured and essentially rule-based, which may explain why frameworks like this are so fragile. Anyhow, the framework itself won't be any surprise to people with learning design experience, though they'd probably be disappointed by the short one-paragraph description of 'learning opportunities' as 'rehearsals for assessment.'.

Not long after the initial popularity of ChatGPT David Wiley argued that the new form of open educational resources would be found in the form of 'open prompts'. This new open access book (391 page PDF) feels like the implementation of that idea. The fourteen chapters each take an 'evidence-based' learning theory or pedagogical strategy, ranging from 'think-pair-share' to Gagne's 'nine events' to 'science of learning'. They outline the strategy, develop a set of prompts that implement the strategy, test the prompts, and report the outcome. The papers are very lightly edited (if at all) and there's a lot of variability in length and quality, though I wouldn't describe any of them as bad. A little more care in presentation would have helped a lot (things like numbering the chapters, sizing the images properly, nicer heading font selection, etc (though these may be limitations of the PressBooks format). Also, there's no mention anywhere of newer developments in the field, like agents (except for a mention in the section on Bloom's Taxonomy), model context protocol (MCP) and the other tools AI models can access.

UNESCO has released a short (9 page PDF) 'programme and meeting document' describing open educational resources and explaining why people might want to adopt them. The majority of the document focuses on questions people might have, for example, "I want to keep some control over my works" or "I do not want others to judge the quality of my educational resources."

The headline really distracts from the core question and I wish people would stop presenting arguments of this form: "X alone won't fix Y". No kidding. We live in a complex interconnected world. Nothing 'alone' fixes anything. The real questions concern what things are necessary (if any) and what combination of things will be sufficient. We also need a good definition of what it is exactly that needs fixing. Now consider the summary of this article: "Even sophisticated AI systems need human support and institutional capacity to succeed in addressing social problems." What we need to learn, really, is what kind of institutional support and institutional capacity are required. At one point we read it's "adequate infrastructure, dedicated administrative support, and enabling leadership." So - the usual.