Stephen Downes

This is a topic we touched on a lot when we were defining connectivism back in the early 2000s. Connectivism was in part the thesis that the way individuals - and societies - learn is described by graph learning. Graph learning is explained mathematically with graph theory, with origins in Euler's Seven Bridges of Königsberg problem, and realized physically in any number of naturally occurring graphs, including the chirping of crickets, bird murmurations, metronomes on a plank, and social networks. Graph neural networks formed the basis for deep learning, which in short order became the AI models we know and love today. The next challenge, which is alluded to at the end of this article, is "ask how can we best integrate graph structured data with artificial intelligence (AI) to allow for the encoding of graphs for large language models (LLMs)?" Via Data Science Weekly.

This is a fun article but if you want to go meta, read the comments. Tne idea is that you can put your full effort into play, but only up to a certain point, at which it stops being a game and becomes (one of) serious, dangerous, harmful or banned. The author - known only as Hastings - recounts a number of such instances through their childhood. The twist, that comes at the end, is: "Anthropic appears, in their latest report on usage in education," seems to be sensing these limits. Ah. But as I say, do read the comments. Via Data Science Weekly.

In the database world basic operations are abbreviated CRUD: Create, Read, Update, Delete. This article basically does the equivalent for transactions, identifying the four basic operations, which are: Execute, Order (as in a time sequence), Validate, and Persist. No acronym. This article explains each, provides some examples, and then looks at a number of transactional databases from this perspective: FoundationDB, Google's Spanner, TAPIR, Calvin, Consistent Unordered Replication Protocol (CURP), and TicToc. Fun question: "Draw any made up diagram of a possible ordering or interleaving of execute, order, validate, and persist. Now answer the question: how would I need to design a database such that it would decompose to this diagram?" Via Data Science Weekly.

You can just hear Alex Usher gushing as he interviews s Nicolas Badré, Chief Operating Officer of the Galileo Group, a private company that operates "70 prestigious schools across 20 countries and 120 campuses." Usher asks, "why would so many people choose to pay for education when they don't have to?" Badré responds by talking about employability, student experience, and a "values-based dimension" focusing on "innovation, entrepreneurship, and high standards." It's the same sort of answer we get from private hospitals in the U.S. My (more honest) response to such a question would be: marketing, networking and elitism. It's the creation of a two-tier system to confer power and prestige. It's people paying for a Hermès education (also available in France) and the way the handbag becomes a social cue. Plus: Usher also features prominently in a University Affairs article where he notes, correctly, that "It has become starkly apparent that many critics of EDI in the United States aren't just opposed to diversity initiatives — they're opposed to civil rights. There is no presumption of good faith in these debates." But it's the challenge of our age, isn't it: how can you maintain support for things like equity, diversity and inclusion when you undermine institutions, like public education?

This article gets at some of my own frustrations with the educriticoblogosphere. Jasmine Sun writes, "analysis around how AI will fit into our human world remains rudimentary... many of the strongest thinkers about society and politics - the people I want most guiding diffusion - continue to believe that LLMs are buggy auto-complete." It's not that, of course. As Sun comments, "AI discovered wholly new proteins before it could count the 'r's in 'strawberry', which makes it neither vaporware nor a demigod but a secret third thing." See the illustration. But "Many tech critics have become so terrified of 'critihype' that they've abdicated their actual job, which is to anticipate and examine how frontier tech may change society in ways good and bad. And yes, that requires some speculation about things that haven't happened yet. Hype-busting is useful but cannot be the whole game!"

Good post making the following point and then expanding on it: "Don't let anyone tell you different: there is no such thing as a separate 'infrastructure phase' in technology. The tech industry loves our neat narratives - first comes infrastructure, then applications follow. It's tidy, linear, and completely wrong." I would add: what we are learning today - not just in tech, but across the board - is becoming tomorrow's infrastructure. We'll still need a few experts to build that infrastructure, but mostly we don't need that sort of work any more. My first job in tech was 'computer operator' - imagine! We needed people specifically to operate computers. Elevator operators, typists, switchboard operators - gone. No, now we need people to do the new applications - but they're evolving as fast as the infrastructure, alongside it. Translators, copy editors, illustrators - no longer needed. Even my current job - programmer, researcher, journalist, educator - is being replaced. Becoming part of the infrastructure