2024 week 39

⚙️ Work

Lots of meetings this week, but the most fun one was planning deeper collaboration with AI and emerging technology initiatives in our faculty of eduction.

And we had a great session for TI staff to discuss how we are working toward Truth and Reconciliation. Christine shared some sobering stories that underline just how important and unfinished this work is. And this was one week after Calgary Police killed John Wells for loitering-while-Indigenous in a hotel lobby (coincidentally, our Reconciliation session¹ was held just a few hours before preliminary details of the ASIRT investigation were made public).

I met with a (very) senior faculty member this week, and they asked me about my plans for the long weekend. I paused for a second before responding. It’s not a holiday, and not a long weekend. The university is closed so that we can all actively engage with events as part of the National Day for Truth and Reconciliation. They said they had no idea, and were looking forward to the long weekend.

🔗 Links

Academic Innovation

• Mike Gillespie @ The evolllution²: Creating Emerging Technology Sandbox Spaces for Faculty

  Saskatchewan Polytechnic launched a Technology-Enhanced Learning Plan in 2021. As part of the implementation, they created a set of “sandbox” spaces for instructors to explore various emerging technologies.

  From their TEL Plan:

  Create a Learning and Teaching sandbox space at each campus for faculty and students to explore and experiment with educational technologies.

  These are all in-person technologies like VR and 3D printers and drones, etc., but the model should work for projects at other institutions like OpenETC for online sandboxes.

Course Design

• Tim Klapdor: Learning Types - sharing some of his course design work at the University of Adelaide. (via Tim’s mastodon feed)

  This is interesting work - and not to be confused with Learning Styles. These are types of learning that can be intentionally designed into courses. He also talks about “learning patterns”, which looks like a nice way to talk about types of activities (and also connects to some of the findings in my dissertation).

  The resources are a good starting point, but I’d love to see them built out with examples in different disciplines - what does this stuff look like? And, how can these types of learning be authentically assessed?

AI

• Google’s NotebookLM AI tool (via Clint Lalonde)

  I fed my dissertation into NotebookLM, and was gobsmacked at the podcast episode it produced.

  Also, in the 18 months since I finished my dissertation, 115 people have readdownloaded it. In the few days since I shared the .mp3 of the podcast episode, it’s reached ~359 people³. Sure, it’s incomplete and simplifies the document and misses 99% of it. But ~359 people have a basic idea of the premise of my work, who wouldn’t have (didn’t) read the document anyway.

  And, you say, of course ~359 people accessed the podcast thing - I shared it on my blog, on Mastodon, LinkedIn, and on Teams. But that’s exactly how I’ve shared my dissertation. There’s something going on here - either “this is new and I want to see what it’s about” or “ok - that’s an easy way to learn about this work”. (likely much more of the former than the latter, I’m guessing, but still…)

I’ve been following the 2001: A Faster Odyssey mastodon feed. This frame perfectly sums up Generative AI in 2024.

Keir Dullea as Dr. David Bowman, after dealing with an overhyped AI and staring into the abyss, in 2001: A Space Odyssey (1968).

2 links from Lucas Wright @ UBC’s Digital Tattoo initiative:

• Your Privacy and AI Tools
• Generative AI and Bias

VR and AI and simulations

• Teixeira, L., et al. (2024). Virtual reality with artificial intelligence-led scenarios in nursing education: a project evaluation. British Journal of Nursing, 33 (17). https://doi.org/10.12968/bjon.2024.0055 (EZProxy link)

  VR is a useful tool for simulating interactions with patients. Adding an AI interface sounded promising, but it looks like the students spent time trying to rephrase their questions to get the AI to understand them (which they’d likely do with humans, but it wouldn’t feel like ‘prompt engineering’ as much as ‘wanting someone to understand what you’re saying’). But they found (with a very small sample size) that the voice interaction with the simulated patient did enhance the experience for students and also provided an opportunity to practice communication skills. In comparing the conventional menu-based non-AI interface with this natural-language voice interface:

  The conventional menu-based interaction method gives students more structure and guidance, which was also perceived as positive but more relevant for the early stages of their pre-registration programme and development as a student nurse.

  So the structure and scaffolding of a more controlled interface might help with novice learners, while the more fluid and organic nature of a natural-language interface might help with more advanced learners?

Doppelgangers

• Danny Liu @ Teaching@Sydney: How Sydney educators are building ‘AI doubles’ of themselves to help their students

  They’re using a new platform called Cogniti.ai (built by “a team from the University of Sydney”⁴). There isn’t a lot of info about its relationship with USydney, but it is mentioned in their latest report to the Australian senate.

  Donald Sutherland as Matthew Bennell, identifying AI Doubles posing as educators, in Invasion of the Bodysnatchers (1978).

AI and Research

2 links via Ethan Mollick (shared on LinkedIn):

• Han, S. (2024). Mining Causality: AI-Assisted Search for Instrumental Variables. arXiv preprint arXiv:2409.14202.
• Si, C., Yang, D., & Hashimoto, T. (2024). Can LLMs Generate Novel Research Ideas? A Large-Scale Human Study with 100+ NLP Researchers. arXiv preprint arXiv:2409.04109.

They’re both pre-prints of articles, so who knows what might change as they finish the peer review process. But. “But GenAI is strictly reductive and can’t create anything” seems to matter less every day…

This is awesome / we’re doomed. (or somewhere in between)

Tech

Tim Owens is trying out the Meta Ray Bans to record video of his repair work at Reclaim Arcade. Amazing first-person video, without having to strap a GoPro to your head.

📚 Reading

I finally finished reading The Land of Painted Caves - the final book (of 6) in Jean M. Auel’s Earth’s Children series, exploring fictionalized accounts of prehistoric peoples in Europe. Auel started writing the series in 1977, so the science is a bit off. But it provided an entertaining account of how cro-magnon and neanderthal peoples may have interacted, and how innovations may have spread across Europe.

Reading the series, about half of it felt like reading an encyclopedia built from the author’s notes from taking classes and workshops and tours. About 10% of it was similarly encyclopedic detail applied to romance novelization of the lead characters, cro-magnon throbbing manhood and the like. The rest was pretty interesting. I stuck with the series based on a recommendation from a colleague. I’d have stopped after book 2 otherwise. Interesting, but there are much better uses of the time - it soaked up 5 months worth of fiction reading for me.

🧺 Other

Blog

Despite just feeling right, the departure typeface’s cleanly rendered 11px size was just too small for my old eyes. Increasing it just made it look blockier and blurrier. Back to the Jetbrains Mono typeface for now.

🗓️ Focus for next week

• Monday is National Day for Truth and Reconciliation. UCalgary is closed for the day, and I’ll come to campus for a talk by Michelle Good.
• Meetings.
• TI Leadership Development session - the leadership team is coming together for a series of development sessions, working with Jolene Ondrik