Stephen Downes

A person is given a task - building a bridge, say, or writing a report. What do they do? They could just hire a contractor or enlist AI, but will that produce the best result? How do they know? This, to me, is the grist of assessment in the AI era. It's not - as this report (37 page PDF) proposes - 'AI-Resilience', which is "design (that) ensures that core learning outcomes cannot be easily outsourced to AI... by thoughtfully creating conditions and structures that make it hard for students to use AI to complete the core." It is (to coin a phrase) 'AI-Agnosticism', which is evaluation design that sets a genuine tasks and doesn't care how it accomplished, only that it is accomplished as well as possible. What's fun here is that this sort of design is hard for an instructor to evaluate, but much easier for an AI, which can compare and consider multiple ways of accomplishing the task. 

Nik Bear Brown writes by email, "I wanted to share a project I've been working on called "Lyrical Literacy" as part of my work with Humanitarians AI. See the video here.... (that) points out that singing engages multiple brain regions simultaneously - language centers, memory networks, emotional processing, and motor control systems all working together. Despite the scientific evidence of music's cognitive benefits, it's unfortunate that singing programs are often first to be cut from schools." I did like the video, partially because of the message, partially because of the music.

In anticipation of the (very real) Jacinda Ardern memoir A Different Kind of Power, this article reviews several AI-authored alternatives from 'Leading Quietly' to 'Power with Kind Purpose'. Unsurprisingly, they do not fare well (though one gets a rating of "9 out of 10 tear-sodden tissues"). Obviously, none of them could stand against the real thing. Perhaps when AI tools improve they may serve just as well, when there isn't a genuine memoir available (or, in some cases, to stand against an obviously self-serving memoir). But the practice Amazon is perpetrating - allowing these titles to try to fool the reader in to buying one thing when they intended to but another - is underhanded. It's the sort of thing an unethical corporation would do. Via Kate Bowles.

I've made similar arguments in the past, but again, you don't have to listen to me, listen to other people in the field. In this case, the author debunks the viral "bottle of water per email" ChatGPT claim propagated in media such as the Washington Post. "LLMs replying to users simply do not use up that much water. Some water is sometimes used for cooling, but this is negligible. Most of the water attributed to LLMs is used up for generating power, because power generation requires water. Querying an LLM generally uses up less power, and therefore less water, than making toast or leaving one of your lights on for a few minutes."

This article introduces the idea of mixing AI and design tools. It's a good mix, in my view. I used to create large diagrams to map out ideas (usually associated with the history of philosophy). I'm not sure AI would help me learn this subject the way creating diagrams, but it would certainly help me create the end product. The article discusses two free open source diagramming tools, draw.io (which I have used) and mxGraph (which I haven't) and shows how to combine them with AI services to automatically generate the diagrams. There's a lot of good original thinking here.

Not that I have time to even begin enjoying this 20 part series, but I pass it along in the hope that readers fine it useful. Via George Siemens.