Web Neural Network API
W3C,
2023/04/17
The W3C is looking for implementations of the Web Neural Network API. "The Web Neural Network API defines a web-friendly hardware-agnostic abstraction layer that makes use of Machine Learning capabilities of operating systems and underlying hardware platforms without being tied to platform-specific capabilities." Sample application areas include person detection, semantic segmentation, style transfer, super resolution, captioning, summarization, and much more. The idea is that the AI would be built in to our devices, and then accessed by applications and services. The concept has been around since 2017 and published as a W3C working draft in 2021 (more) but obviously obtains a new level of importance today.
Web: [Direct Link] [This Post]
Of Accountability and Interdependence
Ann Gagné,
All Things Pedagogical,
2023/04/17
Accountability is one of those tough issues in education and raises issues of "who we are accountable to in our day-to-day lives and who should be accountable to us." It's especially relevant, writes Ann Gagné, at the end of the semester and assignments are due. It also comes up in the context of things like collaborative projects, when students are (we are told) accountable to each other. In this post, accountability is depicted as being related to one's level of education ("how much someone actually understands, considers, and values interdependence") and ethics ("the most grounded, community informed, social justice aware spaces are the ones where folk see themselves as accountable to others"). But I think the concept is an often convenient bludgeon. After all, there's a world of difference between accountability we freely choose (such as entering into a marriage or contract), accountability that just arises out of nature (such as parenting a child), and accountability that is imposed on us (by a government or authority). Image: Rhythm Systems.
Web: [Direct Link] [This Post]
What is a knowledge-rich curriculum? Principle and Practice.
Tom Sherrington,
Teacherhead,
2023/04/17
So what is Twitter serving up as 'education' or 'learning theory' related content these days (that is, as recommendations based on the content of my @OLDaily feed)? This sort of stuff: "the specifics of what we want students to learn matter and the traditions of subject disciplines are respected. Skills and understanding are seen as forms of knowledge and it is understood that there are no real generic skills that can be taught outside of specific knowledge domains. Acquiring powerful knowledge is seen as an end itself; there is a belief that we are all empowered through knowing things and that this cannot be left to chance. There is also a sense that the creative, 'rounded and grounded' citizens we all want to develop – with a host of strong character traits – will emerge through being immersed in a knowledge-rich curriculum." Think of this as the hydroxychloroquine of education policy. It's not simply that I think this is a mistaken view of learning and development, it's that this view is aligned very closely with a particular political perspective, and serves the need of that perspective more than it reflects what learning actually is and what learners actually need.
Web: [Direct Link] [This Post]
When is a Concept A Priori?
Emmanuel Ordóñez Angulo,
PhilPapers,
2023/04/17
Some things, we are told, just mean what they mean. The concept of a 'square', for example (four sides of equal length at right angles) doesn't depend on our actual experience of a square. It just is what it is. But what about more concrete concepts, like, say, ethics? People objected to Philippa Foot's argument that 'the grounding of a moral argument is ultimately in facts about human life' because, they said, "moral knowledge, like logical and mathematical knowledge, (is) supposed to be a priori." What about taxonomies and natural kinds, like 'human'? Isn't a 'human' a human by definition? I think that like Foot's critics we take a lot of supposedly a priori concepts for granted, when really, they reflect experiences, culture, or nature. That's what this paper (17 page PDF) discusses.
Web: [Direct Link] [This Post]
A look at open-source alternatives to ChatGPT
Ben Dickson,
TechTalks,
2023/04/17
This article profiles five open source alternatives to chatGPT: LLaMA, Alpaca, Vicuna, Dolly, and OpenAssistant. "The recent push to bring open-source LLMs has done a lot to revive the promise of collaborative efforts and shared power that was the original promise of the internet," writes Ben Dickson. In related AI news: an AI-generated Drake x The Weeknd tune hits the charts (and gets taken down (but here it is)). Drake is not happy. Meanwhile, "the winner of a major photography award has refused his prize after revealing his work was in fact an AI creation." Via SAIL, here's a glossary of AI terms in education. Also, a guide to understanding LLMs (some good information here). And finally, Re-memory: scam or soulmate?
Some reports about AI in education: now there's more room for child led, open- ended 'CLOE' time at schools. "More personal, social, creative, and critical thinking skills really need to be learned through direct experience and trial and error." And for the complete opposite of this, Brainly is announcing "new AI functions allowing Learners to 'Simplify' or 'Expand' answers." Class Technologies is announcing "plans for a beta release of A.I. Teaching Assistant." Also, there are suggestions that "many Chinese have begun using ChatGPT to do their ideology homework." The 74 says schools must embrace chatGPT.
Web: [Direct Link] [This Post]
Hegemonic Design Bias in Massive Open Online Courses (MOOCs): A Conceptual Framework for Why MOOCs Struggle to Democratise Learning
Michael Meaney,
EdArXiv,
2023/04/17
This paper (22 page PDF) argues that "the design and iteration of MOOCs reflect the preferences and reproduces the advantages of an already elite group; namely, learners with a college degree." Now the paper discusses only "Coursera- and edX-style xMOOCs produced in the United States (USA) predominantly in English" so perhaps we could hardly have expected different results. Still, it's an interesting exploration. However, the paper is not well-written. The author struggles with style and clarity, and frequently assigns agency incorrectly (for example, saying things like "MOOCs in this paper refer to Coursera- and edX-style xMOOCs..." when it's the paper that refers to edX-style MOOCs, not the MOOCs; even the title is unclear, using the word 'for' instead of a verb, and suggesting that MOOCs, rather than their creators, 'struggle' to do something). I recommend that all writers study effective writing principles (this, interestingly, is precisely the sort of background underrepresented learners are lacking and not provided in most traditional MOOCs).
Web: [Direct Link] [This Post]
Growing the Fediverse
Johannes Ernst,
reb00ted,
2023/04/17
Johannes Ernst points to two key improvements needed in the fediverse: "seamless single-sign-on across the Fediverse apps that I'm using" and "my social network should come with me when I use different Fediverse apps." It goes without saying that neither of these should be centralized services (but I'll say it anyway).
Web: [Direct Link] [This Post]
There are many ways to read OLDaily; pick whatever works best for you:
This newsletter is sent only at the request of subscribers. If you would like to unsubscribe, Click here.
Know a friend who might enjoy this newsletter? Feel free to forward OLDaily to your colleagues. If you received this issue from a friend and would like a free subscription of your own, you can join our mailing list. Click here to subscribe.
Copyright 2023 Stephen Downes Contact: stephen@downes.ca
This work is licensed under a Creative Commons License.