OLDaily

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

Where did germane cognitive load go? - by Greg Ashman
Greg Ashman, Filling the Pail, 2022/11/25

---

This article points to two major issues with cognitive load theory. First, "germane cognitive load makes cognitive load theory unfalsifiable... (but) Unless a theory can make predictions that can be tested as right or wrong, it is an unscientific theory." Second, the theory can't really "explain the potentially conflicting experimental results where both raising and lowering cognitive load can lead to better outcomes under different circumstances." That (to me) is why the advice coming from cognitive load theorists to always lower cognitive load sounds so unintuitive.

Web: [Direct Link] [This Post]

A bot that watched 70,000 hours of Minecraft videos could unlock AI's next big thing
Will Douglas Heaven, MIT Technology Review, 2022/11/25

---

This AI project uses online videos as "a vast and untapped source of training data" to help systems learn how to perform tasks in Minecraft by watching and imitating human Minecraft players. It's interesting to see the parallel between machine learning and human learning, as this project depends in a way on a process of scaffolding to make the task achievable: "The team's approach, called Video Pre-Training (VPT), gets around the bottleneck in imitation learning by training another neural network to label videos automatically." (I can't tell any more whether MIT Technology Review is using paywalls, so if you hit one, please let me know).

Web: [Direct Link] [This Post]

Cognitive Science and the Different Kinds of Computation
Gualtiero Piccinini, The Brains Blog, 2022/11/25

---

This is one of my core messages about cognitive science as well, though Gualtiero Piccinini is probably more qualified to make the point. He writes, " In a 2013 paper with biophysicist Sonya Bahar, I have argued that neural computation, in general, is not digital; therefore, theories of cognition need to be formulated in terms of (what we know about or expect to be possible via) neural computation."

Web: [Direct Link] [This Post]

Researchers Find Stable Diffusion Amplifies Stereotypes
Justin Hendrix, Tech Policy Press, 2022/11/25

---

This article points to how "current methods to mitigate these effects fail to prevent images perpetuating racist, misogynist and otherwise problematic stereotypes" produced by a Stable Diffusion AI. "œI think it'™s a data problem, it's a model problem, but it's also like a human problem that people are going in the direction of 'more data, bigger models, faster, faster, faster,' Hugging Face'™s Luccioni told Gizmodo." Or as a speaker at the conference said today, "AI is 98% human." In an email Alan Levine looked at the depictions of teachers. "Notice the teacher tropes, either bookshelves or chalkboards." And compared to DALL-e: "Oh my. DALL-E only places teachers in front of chalkboards! But worse, for intellectual (left) the results are weighted to balding white males in glasses." Here's the project page.

Web: [Direct Link] [This Post]