Stephen Downes

The article in a nutshell: "the key question educators face right now is this: Are we using AI to enhance learning, or to replace some learning steps that turn out to be essential?" This is a good discussion of the two approaches. I would not that it's not just a question about AI, it's a question about teaching method. Do you focus just on getting students to be able to give the right answer, or do you focus on the work and practice that enables them to actually use this knowledge?

This paper walks the reader step by step through the basic concepts of natural language processing (NLP), expanding to large language models and terms like 'perplexity', and through to a look at what to expect in the future. "The vast majority of architectures are neural networks (sometimes called artificial neural networks
to differentiate them from biological ones)... A general observation about neural network architectures (but not a necessary or defining property) is that the relationship between their numerical calculations and the task-solving behavior of a model (after its parameters are set) is not explainable to human observers."

This website looks super-temporary so download the content while you can. It's a set of questions and answers ablout natural language processing (NLP). It's comprehensive, and it's informative, and I imagine its primary use will be by hiring managers trying to figure out what to ask candidates. It was authored by a second year computing student who, it appears, prefers to remain anonymous. It's really informative and you will know more after reading it. Also, it made me think, what of you went through it point by point constructed a 'learning theory' out of it. Call it 'deep connectivism' or something.

We may all remember the digital Sheldon, but would it actually work in practice? This article studies whether a telepresence robot would be effective in a teaching setting. "The aim was to analyze students' perceptions of their teaching experience with a robot, to verify whether it would be possible to interact with students during the experience, and to determine the advantages and challenges of this teaching method." Published in French, but these days easily translated into English.

Participants in this study used the Lego Mindstorms EV3 Education robots to develop skills in computational thinking. We already know educational robotics is recognized is "an effective tool for enhancing students' computational thinking skills... because students must consider the robot design and behaviour programming that will best interact with the physical world within the context of a given situation or problem." This study measured the students' own perceptions of their abilities.

I haven't tried Talkyard as an alternative to Disqus for hosting comments, but this article describes his search and decision, and I have enough confidence in the author to pass it along. I'm sometimes tempted to reinstate comments for OLDaily, but I never did get enough comments for it to be worth spending much effort on (my feeds on Bluesky and Mastodon draw the occasional remark, but that's it).