OLDaily

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

The Talking of the Bot with Itself: Language Models for Inner Speech
Cameron Buckner, PhilSciArchive, 2025/01/06

---

In a sense, neural networks have always been able to 'talk to themselves'. There is back propagation, for example, where feedback flows back through the layers of a neural network, correcting weights as it does. Or there are recurrent neural networks, where neural output is saved to become neural input, creating in effect cognitive loops. But 'talking to ourselves' or the idea of an 'inner voice' has always been thought to be something more abstract, definable only in terms of lexicons and syntax, like a formal system. This article (34 page PDF) grapples with the idea, considering it from a conceptual, theoretical and then practical perspective, running us through Smolensky's argument against Fodor and Pylyshyn through to things like the 'Inner Monologue Agent' from Google Robotics and Colas's language enhanced 'autotelic agent architecture'. "Instead of viewing LLMs like ChatGPTs as general intelligences themselves, we should perhaps view them as crucial components of general intelligences, with the LLMs playing roles attributed to inner speech in traditional accounts in philosophy and psychology."

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

I was lost in the cesspit of social media. Then Jane Austen showed me the way out
Joel Snape, The Guardian, 2025/01/06

---

When I closed the last page of Gibbon, it felt like saying good-bye to an old friend. The same with Tolstoy. But it's not just the medium. Lawrence of Arabia - the movie - is a regular visitor to my recollections. The Expanse similarly. Don't be fooled - books aren't inherently good, social media isn't inherently bad. This could be Austin, or it could be Twitter: "the asides are savage, and the social critiques are delivered in a flurry of bon mots." I did not feel elevated after reading Pride and Prejudice; if anything, I felt debased, as through I had spent time with an old gossip rag, or indulged in an evening with the algorithm. The problem with social media isn't that it's bad media, it's that so many on it do media so badly. 

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

More Eating the Future
Alex Usher, HESA, 2025/01/06

---

Alex Usher doubles down on his abhorrent 'eating the future' argument, but he also makes a good point that I'll focus on here: "There is a really basic argument about the value of postsecondary education which somehow, postsecondary institutions are losing with governments and, I think by implication, the public. That, and nothing else, needs to be the focus of institutional efforts on external relations." By framing the argument as being between 'postsecondary education' and 'pensions' Usher is not following his own advice, obviously. I don't think attacking old people will get the sector where it wants to go, Usher's biases notwithstanding. At the same time, if it wants to remain robust, postsecondary will have to find a way to appeal to the broader public. 

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

Micro-learning in 2025
David Truss, Daily-Ink by David Truss, 2025/01/06

---

Dave Truss describes how microlearning has changed over the years (where 'microlearning' is thought of as 'getting an answer to a question'). We used to try the encyclopedia (or maybe, if it's me, the World Almanac or CRC Handbook of Chemistry and Physics). Later, we would check Google or Wikipedia. Later still, for some people at least (probably not me), we'd ask social media. For some things, we'd ask YouTube. Today, says Truss, we're just as likely to as an AI. "AI will do two things for you. First it will curate your learning for you. And secondly it will be adaptive to your learning needs." Except - right now, being adaptive to my learning needs is something AI isn't particularly good at. Though this will change. Truss argues that this will have an impact in classrooms. Maybe. But AI will be as welcome in classrooms as the plague. I mean, they're banning phones. Students will be the last to use AI as part of what they do, not the first.

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

Bluesky Won't Save Us
Katherine Alejandra Cross, Liberal Currents, 2025/01/06

---

This is a good article that takes an opinionated look at what might be called the inherent contradiction in distributed social media. It's this: "The problem with 'Power to the Users' as an ethos is that there are simply some things that users can't do. Foremost among them is content moderation at the scale of millions of users." You get "versions of Stormfront or Gab that no one can do anything about." The thing is, if the platform is owned and run by a large corporation, as Bluesky is, then there is the expectation that the owners can do something about the offensive and potentially dangerous content. It's a sort of centralized decentralization, and turns the owner into a sort of global Supreme Court - something I don't think Bluesky wants to be. Via Laurens Hof.

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

Pattern Recognition: How can AI support learning and are we even organized for that?
Mark Oehlert, Relentlessly Curious and Deeply Engaged, 2025/01/06

---

I read Pattern Recognition when it came out and it has been kind of a secret weapon for me for the last 20 years (more, if we include William Gibson's earlier works). It's a good premise for Mark Oehlert to use to launch this article on reorganizing learning (and companies) of the future. For example, he quotes a recent (paywalled) paper that reports "In our study, the algorithms couldn't fully optimize because the org chart kept decision-making locked in silos. Once those constraints were lifted, AI delivered far better results—spotting trends and opportunities no single team could see on its own." Now, ChatGPT has a tendency to do a lot of the stuff Oehlert describes - telling me 'why it matters' and 'how to implement' and other such stuff that might be useful (but often isn't, and just pumps up the token count). We need output in a form we can use - again, not 'tutoring' mode, but performance support mode -  'AI-in-the-Loop' mode.

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

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