Ed/Tech must-reads 190825

AI 4 ID community from NRZ Malik

It’s hard to open a webpage these days without someone having a bunch of knowledge to share (or sell) about how to use GenAI but I happened across this online community about AI targeted at Instructional/Learning Designers the other day and I have to admit, it seems to have a decent amount of useful content. Resources include tool updates, design workflows, eLearning authoring tools, a prompt library and more. It’s hard to say whether the host of the site is just obsessive, as basically all of the trending posts and recent activities seem to be from him but hopefully that’s just a good reason to join and contribute.

Assessment integrity in the age of AI: Where to from here? from Needed Now in Learning and Teaching

Miriam Reynoldson (RMIT) has been one of the more vocal GenAI refuseniks in my socmed feed of late but her passion for good learning and teaching and student equity shines through her work - even when I don’t agree with all of her positions. This contribution to Needed Now (always a worthy read) goes into some depth about the perennial AI in AI issue and provides some long-awaited alternative approaches which take us beyond ‘here’s why everything is terrible’. She offers a detailed and thoughtful overview of the current challenges and some sweeping solutions for making education better in the long term.

A comprehensive taxonomy of hallucinations in Large Language Models from arXiv

I think it is safe to say that we have all been simultaneously amused and horrified by the profoundly bizarre hallucinations from various GenAI platforms - from the suggestion that we should eat rocks every day as a little treat to the absolute certainty that there are 3 bs in the word blueberry. This pre-print paper from Manuel Cossio goes into far more detail about the types and causes of these hallucinations than I would have considered possible and even just a casual scan of this article is enough to firm my sense that we still aren’t quite there yet.

Reflections in Ink: Capturing a HERDSA 2025 keynote through comic panels from HERDSA Connect

I’ve long appreciated the idea that we can process the torrents of information in conferences by re-presenting them in other formats. Diana Saragi Turnip turned her hand to this (with AI) during auditing SocMed maven Amanda White (OAM)’s recent keynote at the HERDSA conference. Using a manga comic style, she captures key ideas from the talk about influencing learners through presence. Well worth a look

Avoid These 3 Common Mistakes When Choosing an LMS from eWorks

While this blog post is a little bit of an ad for their services, eWorks has been doing good things as part of the VET sector for many years (anyone remember Flexible Learning Toolboxes?) and the points they raise about LMS selection are well made. They emphasise the need to remember scalability, the data that a platform can provide and the importance of fit over price.