Matt Bury

From reading the comments here, it'd appear that the phenomenon of departments spending almost all the online & distance education budget on hardware, software, & resources, & only a tiny fraction on teacher CPD to enable them to use it is still a thing.

I sometimes give CPD sessions to teachers on how to use the IT resources available to them. Before the sessions, teachers typically give the impression that they're already familiar with at least some of it & one could be forgiven for assuming that they know what it is & how to use it. However, during the sessions, from the intensive note-taking & eager questions, it frequently turns out that they're pretty much novices in this domain & hadn't realised what the practical applications are.

Then comes the issue of having the time & energy to do the work of implementing what they've learned in their own courses, which can be considerable.

I'm sure all of us here find implementing technology enhanced learning (TEL) second nature or close to it. In my opinion & from my experience, it's a steep learning curve for most teachers, many of whom went into teaching to work with people rather than computers. I think any TEL adoption by any department really needs to take this into consideration. The teachers & department heads themselves that I've come across are typically fully aware of this but they're not the ones who make the executive decisions & sign the cheques for the nice EdTech sales reps, rather, it's foisted upon them on top of already heavy workloads.

Just my €0.02!

Yes, for particular levels of language/communicative proficiency, I've found it doesn't respond well in the initial prompt to generate a text but it you tell it to rewrite it at a level, it does so fairly well. However, it doesn't seem to know how to do CEFR levels so I use US grade levels 1-12 instead, i.e. "Rewrite [the text] at grade 5 level."

I've tried it for various uses in English language teaching. It's great at generating example texts in whatever genre on whatever topic with whatever arguments and attitudes you ask of it. The more context and profile info you give it, the more authentic and distinctive the output texts are.

It's also good at listing main points, talking points, etc., around particular topics.

However, the quizzes that it generates are very hit and miss, and in some cases editing would take longer than writing it yourself from scratch, e.g. sometimes MCQ distractors obviously have very low efficiency and some questions are on surface features of the topic or irrelevant.

Also, it's analytical and critical responses on students' writing is particularly poor and generic no matter how much context and prompting I give it. So far, it hasn't been useful for generating feedback on students' work.

For generating dialogic spoken genres, i.e. conversations, it's particularly bad and has a tendency to be overly verbose and to explain and repeat stuff ad-nauseum. I reckon I've coined a new expression, "ChatGPT-splaining" for this.

Still, it's early days and I think as applications are found for LLMs, more appropriate models will get trained so that they give more authentic and useful responses. Let's see what happens!

Since it hasn't been explicitly mentioned, H5P is the default Moodle option for multimedia learning interactions: https://h5p.org/content-types-and-applications It provides activities with a stronger emphasis on visual & multimedia elements & interactivity, which is what a lot of people equate with being more game-like. It's already built into Moodle so you can just start using it. See: https://docs.moodle.org/402/en/H5P Of course, it's free & open source & supported by a number of academic & professional institutions so it's unlikely to disappear or be deprecated any time in the foreseeable future; an important consideration for course design elements.