Stephen Downes

The possibility of closer ties between Canada and Europe is pushing some programs familiar to educators into the national spotlight. One is "Horizon, Europe's research and innovation funding program in which Canada already participates," and is looking to join again in its next iteration. I participated in a Horizon project a few years ago (and if proponents are looking for Canadian partners in the private sector, that's me!). Another is Erasmus+, " the flagship EU program allows people to study and train at institutions across Europe for up to a year (and) has since expanded to include exchanges for work experience, apprenticeships and sport."

Mike Caulfield responds to the argument that "you need to know things in order to learn from an AI." This gets it backward, he says. To succeed means that "as the critics admit, I cannot do that without knowing things which means I have to get on learning things... The system you want in a classroom is one that allows you to amazing things if you buckle down and do the intellectual work." It takes a bit of reading to get to this point - the extended horror movie classification scheme example he uses to introduce the topic doesn't speak to me at all. But yeah - we want challenges that force us to figure it out, and usually, that means a lot of reading, thinking, trial and error, with or without AI. This idea - he grounds it in Vygotski's zone of proximal development (ZPD) - isn't something teachers learn when they learn to teach with AI. Instead they learn about "a system that continues to model these as ways to shovel information into students rather than engines of exploration."

This paper (43 page PDF) gets a lot of mileage out of a study involving only ten faculty as subjects, even if they "represented a rich array of fields." I recognize the value of the "case study, where the case itself is of primary interest rather than serving primarily as an illustration of broader phenomena," but the authors don't limit themselves at all, and the extensive literature review and theoretical focus all reflect a much more traditional social sciences approach, only with just ten people studies. They also feel free to generalize, offering four major "results themes" (paraphrased): shifting from fear to curiosity; ethical clarity as lacking but desired; inclusive design as amplifying equity; and faulty as designers, note gatekeepers. Now these discussions are interesting in their own right, but let's not pretend the 'data collection' provides any evidentiary basis for it. The paper would have been just as interesting, and much stronger, with everything between 'literature review' and 'discussion' elided. Via Doug Holton, who also links to the uncritical AI-generated summary in AIEd Knowledge Base, though honestly, I prefer mine.

I think this list of readings has two purposes. One is to offer a reasonable antidote to the hype and doomerism about AI being offered by big tech and from the media. "It's been stirred up recently, either because of yet another 'whistleblower' or weird Bannon-to-Bernie coalitions, both talking about an AI apocalypse." But it's also serving to converge readers around a set of standard authoritative voices who will define what is relevant and what is not, in the spirit of what Carlo Iacono describes here. As he says, "The expertise is real. It has also been allowed to become jurisdiction over every place the technology enters." I'm not saying we shouldn't read the items listed - we probably should. But we should do so with the understanding that we are being led - herded, really - and that we should still be reading more widely as well.

I am honestly bemused by the fact that the link to the Repository of Remote Laboratories that is the subject of this paper (31 page PDF) is not included anywhere in the text (you can find it if you look at reference 91; I resorted to a Duck Duck Go search). That said, the repository is a genuinely useful resource, with links to 40 remote lab services at institutions and universities around the world. The paper itself is also a useful supplement, containing as it does not only the usual methodology and search sections, but also an analysis of the 40 repositories, classifying them according to implementation challenges and technological advancements. Also from the same issue of iJOE, Hansi Effendi, et al., A Systematic Literature Review of Virtual Laboratories in Electrical Engineering Education (14 page PDF), which concludes, "virtual laboratories in electrical engineering education are effective as complementary learning environments rather than substitutes for physical  laboratories," which strikes me as odd, but may reflect the aforementioned implementation challenges.

Philippa Hardman reports, "Last week, Articulate launched Frontline, its 'biggest product update in history'. The product is positioned as, 'An AI-native product built for teams who create training outside of L&D'." Hardman also documents the responses from instructional designers, which aren't hard to imagine: first, "a sense of being designed out of our own industry"; second, " closer to betrayal"; and third, "a suspicion that the claim itself doesn't hold," that the tool doesn't really have 'instructional design expertise' built in. Hardman considers the third response at length. "The expertise isn't the output, e.g. the course. It's establishing whether there's a performance problem and what's causing it, deciding whether training can fix it." Then she tests Frontline: "What evidence is there that Frontline encodes the right process and decisions, in the way an expert instructional designer would make them?" There is no diagnosis step, she concludes. The design remains the same even when the goal changes. The assessment doesn't match the objective. Frontline does a very small part of the instructional designer's job - "one subsection of one phase" (illustrated).