OLDaily

Welcome to Online Learning Daily, your best source for news and commentary about learning technology, new media, and related topics.

AI learns language through a baby's eyes
James Devitt-NYU, Futurity, 2024/02/08

---

Another thing "AI can't do" that it can do. "Their findings, reported in the journal Science (8 page PDF), showed that the model, or neural network, could, in fact, learn a substantial number of words and concepts using limited slices of what the child experienced. That is, the video only captured about 1% of the child's waking hours, but that was sufficient for genuine language learning." Which should surprise no one because (as we have been saying here for almost a quarter century) learning is the training of a neural network through experience, in both humans and machines.

Web: [Direct Link] [This Post]

What GenAI May Take Away
Sean Michael Morris, 2024/02/08

---

It's interesting how you can tell when people reach the limit of their understanding by the vocabulary they use. Here, Sean Michael Morris describes AI that "floats hypothetical information based upon existing knowledge." This isn't so much wrong as it is meaningless. Vocabulary can also indicate expertise, as for example when a useful distinction is made between 'literacy' and 'mastery', as Morris does here: "Literacy demands a deep understanding of a language, mechanism, technology, endeavour, profession, etc., whereas mastery is an exhibition of a learned set of skills. Literacy does not always lead to using a technology, or ascribing to a set of practices, whereas mastery is pointed directly at adoption." And then there's the middle ground, as when he asks, "Do we need generative AI?" My answer is 'no,' we don't need it." Maybe, but we do need answers to "ongoing challenges from teacher shortages and crowded classrooms to democratizing access to higher education through lower-cost options," and saying we don't need AI means there's some other (as yet undiscovered?) solution at hand.

Web: [Direct Link] [This Post]

Opinions on AI
Matthias Melcher, x28's New Blog, 2024/02/08

---

Matthias Melcher offers an interesting and important perspective on the nature and ethics of AI (and this post should be read in conjunction with Ian O'Byrne's Navigating the "Self" in the Age of AI. He offers a take on such subjects as consciousness, understanding, self and subjectivity in AI systems, and how these (as they must obviously do) have an effect on ethics. A lot of this in informed by and aligns with my own thinking, and a lot of it does not, but it offered me an opportunity to consider and clarify what I would say in response; see this longish thread in Mastodon for my contribution. Image: Saul Dobilas.

Web: [Direct Link] [This Post]

Alignment Assembly on AI and the Commons
Open Future, 2024/02/08

---

As this website reports, "Open Future will host an asynchronous, virtual alignment assembly for the open movement to explore principles and considerations for regulating generative AI." I find the approach fascinating. " We will use the process of an alignment assembly, an experiment in collective deliberation and decision-making. This model is pioneered by the Collective Intelligence Project (CIP)." Here's the CIP White Paper on the process. I agree with the assertion that traditional democratic processes are unable to obtain genuine public sentiment on complex issues, but I worry that circumventing a truly democratic process simply grants excess voice and influence to commercial interests (eg. as suggested by "mixed public goods funding models"). I've signed up and will participate if I can. There needs to be some alignment between an open commons and open AI; simply drafting top-down principles at an in-person 'summit' (as Creative Commons has done) is too high-handed and undemocratic a process to rely upon, even if the rest of us (maybe) get a 'vote' on them some day. Via Paul Stacey. Related: Irving Wladawsky-Berger, Open Source AI: Opportunities and Challenges.

Web: [Direct Link] [This Post]

Acceptance Networks for Self-Sovereign Identity
Phil Windley
, Technometria, 2024/02/08

---

"We can't have broad adoption of verifiable credentials until we find a way to scale their presentation by providing tooling that credential verifiers can use to reduce their risk and gain confidence in the facts presented to them." This summary captures the essence of the article but not the depth of it. Phil Windley explores the mechanisms of both trust (which involves risk) and confidence (which doesn't) and traces the genesis of the latter in acceptance networks. We have such a system for financial transactions, but identity systems aren't there yet. "We have identity providers (IdPs) like Google and Apple who control a closed ecosystem of relying parties (with a lot of overlap)... Fixing this requires the equivalent of an acceptance network for digital identity." This article looks at what's required, the principles that make it possible, and centralized and decentraized models.

Web: [Direct Link] [This Post]

Massive List of Chinese Online Course Platforms in 2024
The Report by Class Central, 2024/02/08

---

By 'massive' the authors mean '24'. Still, the article is useful as an overview of the history and development on online learning platforms in China. "Online education is alive and well in China," they write. "Combined, these platforms offer over 100K online courses."

Web: [Direct Link] [This Post]

SDG 4 scorecard progress report on national benchmarks: focus on teachers
UNESCO, 2024/02/08

---

Sustainable Development Goal (SDG) number 4 is widely known as the 'education' goal. This document (103 page PDF) reports on progress toward that goal, as defined by assessments against a defining set of key performance indicators (KPI) (see the illustration). It's important to note that the report focuses mostly on progress toward these goals, and so reports little progress by countries (like Canada) that have already mostly reached those goals. As for the rest: " progress towards national targets is off track for most indicators. In two cases – the gender gap at the expense of boys in upper secondary education completion and public education expenditure as share of total public expenditure – countries are even moving backwards." However, "progress is faster in the eighth benchmark indicator, school internet connectivity" and "progress is also relatively fast in the percentage of teachers with minimum required qualifications." Via World Education Blog. See also Learning levels unknown for half a billion children.

Web: [Direct Link] [This Post]

We publish six to eight or so short posts every weekday linking to the best, most interesting and most important pieces of content in the field. Read more about what we cover. We also list papers and articles by Stephen Downes and his presentations from around the world.

There are many ways to read OLDaily; pick whatever works best for you:

• Read in your web browser on this web page, updated weekdays
• Receive a daily or weekly email newsletter by subscribing here
• Subscribe to the RSS feed using your favourite feed reader
• Follow @OLDaily@mastodon.social on Mastodon or elsewhere in the fediverse
• Follow @OLDaily on Twitter
• Integrate the open API using the JSON feed