Andrew JenningsRE: https://halo.nu/@theguardian_world_news/116292040529519641
I was such a huge fan of "Soul of a new machine". Brilliant writer.
| aus.social | <a rel="me" href="https://aus.social/@georgegalanis">Mastodon</a> |
Dr George Galanis ☑️
- 35 Followers
- 73 Following
- 237 Posts
Scientist, computer science, electronics, human factors, bouncing radar off the Moon. Aiming for a true distributed internet.
It’s long past time we called out One Nation as a wrecking ball that would destroy the country. With no migrants, this is the future: dying towns, dying industries and no workforce: https://www.japantimes.co.jp/culture/2026/03/23/books/depopulation-japan-books-tom-feiling/ #auspol “Alone in Japan” by Tom Feiling
"Ten things language models will never be able to do (on 40 years in AI)" #AI
https://andrewjennings.substack.com/p/ten-things-language-models-will-never
people again complaining about Uni group assignments with the lazy, uncooperative team member. If ever an experience prepared you for working life, this is it #education (team should probably include a psychopath supervisor for the full experience)
Farmers quick to jump on the media and claim special treatment (“secure the food supply”). 70% of food grown in Australia is for export, most of it by big companies. I’m a bit over them. #auspol
US stance on tariff refunds gives us a look at the future. What do you do when you have the world’s strongest military and a moribund economy? You roam the planet stealing stuff. Until someone stops you. #auspol
John Quiggin<irony on> That would disqualify most One Nation voters, and lots of Libs/Nats. Perhaps we should deport them to wherever they or their ancestors came from <irony off>
https://davidwsilva.substack.com/p/im-sorry-to-burst-your-bubble-you?utm_source=multiple-personal-recommendations-email&utm_medium=email&triedRedirect=true - as he discusses, using AI as a "brainstorming assistant" is actually counter-productive. It will vigorously defend the status quo and tell you that you are wrong, when you are actually being original. #AI
"And if today there is a broad misconception about what AI is and does, that's not a testament to AI's power. It's a testament to how aggressively and relentlessly the hype has been orchestrated. So breathe. And then look closer."
@georgegalanis https://www.mind-war.com/p/rug-pull-musks-martian-myth-falls (in which Musk finally understands your argument)
it's interesting to recall Dreyfus "What Computers Can't Do" and how far we have come. I think if you can express a problem in completely abstract form, such as a lot of programming problems and some mathematical problems, then we are at the point where computers are superior. But there is a vast array of stuff that we are light years away from computers being able to do. #AI #LLM
"LLMs can't find the narrative perspective that unlocks a story. Their own existence—their being-in-the-world, to clumsily borrow a phrase of Heidegger’s used by Dreyfus in his phenomenological work—is too impoverished to think about the text they are producing as a reader. They are immensely capable of organizing large masses of text, but the organization they chose will be the organization that has been most commonly chosen before. It will be inherently perspectiveless, because they have no perspective to add. They can't draw a path that the reader or viewer can follow to see the world as they do. They have no access, in short, to the shape. They can construct language that is fit for purpose. They can take large amounts of information distill it down. In many cases, like software engineering, this is more than enough for extraordinary, even world-changing facility. But they can't produce output that helps you see things from a specific perspective, because the generation of that output is perspectiveless by necessity, and that means that they can't produce outputs that carry the spark of creative identity that makes writing—and visual art, and music—interesting to consume." from : "This was issue #29 of Apperceptive by Sam."