Stephen Downes

Matt Webb is feeling some qualms about the way the fruit fly connectome (ie., the full map of neurons and the connections between them) of the fruit fly is being used for fun and games on the internet. "Do I imagine that the simulated fruit fly is someone being trapped and tormented, again and again and again, in a software prison? No, the fruit fly connectome is unlikely to be sentient. (But) at some point on the scale, sentience starts to be meaningful." Future AI might be based on a kernel of very real connections and weights, something that - perhaps - requires our ethical consideration. Suppose you copied your cat's connectome to create an AI. Suppose you copied your nephew's. Your own.

This post summarizes a new report from Microsoft on AI and Education based on a survey of a few thousand students from selected countries. The report is behind a spamwall but this direct link to the 33 page PDF might work. The summary follows a standard order of presentation - people are using AI, think learning is important, are concerned about some issues in common, and are cautious about the future. There's a section on internationalization (takeaway 3) that relies more on an earlier Ipsos report. My takeaways are a bit different - first, how confident leadership is in its own understanding and how different its perception of AI is from the rest, and second, the table of modes of working with AI: Asking, Delegation, Collaboration, Exploration.

According to this article, "Pressing Prompts is an open collection of teaching activities and resources designed to help higher education instructors engage students in discussions about the social and ethical implications of artificial intelligence." It's divided into topics such as 'Can we trust AI?' and 'Is AI biased?' and doesn't try to answer these questions so much as to structure activities and resources to support student discovery around them. 

I like the ambition of this project though I fear future maintenance because of its dependence on Node. Still, I like how a lot of attention has been paid to ease of installation and startup. "FediHome is a self-hosted, single-user Fediverse home — blog, photos, video, audio, and a live following feed, all on your own domain via ActivityPub. 1.0 marks the project production-ready: a test suite + CI gate every change, the security-audit findings are resolved, the dependency stack is current (Prisma 7, Next 16, Fedify 2), and a public demo runs the exact code at fedihome.social." Here's the source.

This is a good overall summary of an informed perspective on AI in education that avoids the cliches and sees AI as something that is fundamentally new, and not defined by simple metaphors. I appreciate this perspective. Not that I agree with it all - my first reaction on reading the first part was to post this on Mastodon: "Some day in the future philosophers will agree that 'meaning' is a construct invented by logical positivists to play the same role previous (and no longer accepted) constructs like 'noumena' played in earlier metaphysics." Anyhow, Dron also says, "no more cognitive offloading: it's cognitive redistribution," whicch I agree with, and "the answer is not to keep humans in a loop (but) to think of genAIs as being part of the web of people, tools, processes, structures, theories, models, appliances, devices, and other artifacts that extends our cognition," which also makes sense to me.

"Let's stop pretending," writes Hamish Campbell, "we need to be honest, the community side of the Fediverse is a mess. The instance as a community, was a good idea, but just that an idea, it never worked, as the was no code that was needed to build the links that mattered, no mod tools, nothing." The existing fediverse, he says, is built on control, when it should be built on trust. "We need to balance collectivizing resilience. Just like with food, we need shared kitchens, food co-ops, and community gardens – not just individual allotments." I tend to favour a third option - not individualism, not collectivism, but cooperation, based on connections rather than commonality.