If you use Google Chrome and are one of the 30 million using UBlock Origin to block web advertising, you are probably within hours of losing that protection as Google begins to disable the extension. The 'official' reason given by Google is 'security considerations' but the real reason, so far as I can see, is that Google is an advertising company that is losing too much revenue to ad blockers. Brave, a Chrome-based browser, is developing its own ad blocker. But your best bet is to switch today to Firefox (which is the browser I use every day). 

Reddit, whuch is a very popular online discussion platform, has started blocking search engines from scraping user content unless they pay for the right to use their content. As a result, "Google is now the only mainstream search engine that yields recent Reddit results." It's hard not to see this as an instance of commercial media companies carving up the web into proprietary domains. These may be free for now but this won't last; as soon as the open web can be extinguished the commercial web will get very expensive to use.

Cory Doctorow offers some good advice for creators who think they are allied with publishers against AI companies: they're not. A publisher may manage to stop an AI company from using their work to train a language model, but they will not grant the same right to a creator when the company decides to sell the AI rights directly. We've already seen this play out at Reddit and Wiley. 

This is a very technical paper but a lovely read if you're really interested in the deep details of today's transformer-based language models like ChatGPT and others. This to me is what a real science of learning looks like. GPTs are by no means the last word, and indeed, the paper is very careful to identify the shortcomings and limitations (one highlight here is the warning not to trust a LLM when it offers an explanation of its own result, specifically, "the tendency of LMs to produce explanations that are very plausible according to human intuition, but unfaithful to model inner workings"). As an aside, I think there's a PhD dissertation to be had in identifying the mechanisms described in this paper and mapping them to social learning, that is, how people in a society interact in order to help society as a whole learn new things.

The Science of Learning (SoL) is a relatively new name attached to methods associated with direct instruction. Read about it here. It has its critics (and I am one), for example, this argument that "the Science of Learning is misleading when it refers exclusively to cognitive science, memory management, and the brain." The article highlighted here represents an attempt by supporters of the Science of Learning to enlist the EdTech Industry to their cause. It might be a tough sell. As Joel Walsh says succinctly in the Learning Engineering forum, "I can say with certainty that the "learning sciences" people would take major issue with being associated with the "Science of Learning" people, and are often in direct conflict with these people."

Miguel Guhlin focuses on a pretty specific need with this post: knowing what types of files differenct AI chatbots will accept (there's a handy grid) and offering advice on how to convert what you have to a format your favourite chatbot will accept.