Watching TV. #catsofmastodon 🐆
Remote
James Wallbank
@lowtech@tldr.nettime.org
mastodon 4.6.8+glitchThey stole our future. Now we're stealing it back.
62 Followers
101 Following
23 Posts
Joined February 07, 2024
Lowtech Manifesto:
Replying to
@ChrisMayLA6@mastodon.me.uk @books@newsmast.community Umberto Eco remarked that getting books is like buying medicine. You don't wait until you're ill—you stock up, so that it's ready for when you need it.
Looking at their responses in order to assess whether LLMs are self-aware is like looking in a mirror and trying to draw conclusions about what your reflection is thinking.
It doesn't matter what it looks like. It matters how it works.
Replying to
@JulianOliver@mastodon.social The questions that data centres appear to address are posed by capital and capitalists, not by market demand for utility.
"How can capital reshape compute to centralise capacity, creating choke-points in all human collaboration, from culture to commerce?"
"What can we do with simply vast quantities of money (other than set it on fire) when we're out of genuinely innovative ideas?"
"What is the most useful tool for society if everyone's job is like mine—bossing and bullshitting, without real consequence?"
"The internet delivered too much value to users, at too low a price. How can we remake the deal?"
"How can we reshape innovation itself, so that we own the future?"
Replying to
@JulianOliver@mastodon.social Indeed! But imagine yourself sitting on simply vast quantities of capital, that is, effectively, evaporating. What do you do with it? Quick! Solving your problem with one, huge decision is simply easier than solving it with a multitude of smaller deals. And that big deal has the advantage of volatility—you might make big gains with well-timed trades. A load of small deals are much more likely to perform averagely. Yawn!
Replying to on tldr.nettime.org
@w0bb1t@tldr.nettime.org Fantastic! I am thinking about this, and, in time, may respond coherently. However, in the meantime, some instant thoughts:
-
I'll disseminate it via my own, non-public networks.
-
It could be nice to have physical copies.
-
It's interesting to reflect on using "scale" as a way to grasp responses to AI. It presents threats at scales from the individual to the global, and, frequently which scale(s) you attend to when assessing the threat is crucial.
-
If one interprets AI as simply the current dominant pattern of corporate behaviour, its qualities become clearer.
-
What AI does is what AI's for. It's for:
—Centralising Compute —Increasing Energy Usage —Confusing Real and Unreal —Disowning Liability —Raising Money —Centralising Data —Grabbing Attention —Flooding The Zone
Replying to
@mattsheffield@mastodon.social Long response. Sorry, there is a point. I hope it helps.
I'm genuinely surprised by the conceptual weakness of Dawkins's sense of self. For me, consciousness is, at its root, modelling—that is, supporting and adjusting models—worlds in miniature—which we examine to help us predict and respond to the real world. These models are only approximate, but, hopefully, good enough. Simpler animals maintain simpler models.
One of the mysteries of consciousness emerges from the need to place a model of ourselves inside our models of the world. (What would I do if…?) Inside model me there may need to be a model of model me… (How would I feel if…?) Very quickly, the detail disappears: so it's hard to see yourself with any degree of fidelity.
By my understanding, Dawkins is failing to perceive that, to be conscious, you need to maintain a model or models of yourself in the world (and models of yourself modelling the world!), not just to produce words that claim that you do so.
Claude does not maintain a persistent model of itself. Dawkins is mistaking appearance for internal structure, like mistaking a mirror image for a living being, just because it moves.
Replying to
@CuratedHackerNews Brilliant commentary. Friction is good.
I have been horrified by how quickly a self-proclaimedly progressive tech industry has collapsed itself into a value-free amplifier.
Replying to
@gdc Why a new system? Don't vector fonts typically already have anti-collision data embedded within them? I think I recall that each glyph's definition has a high, middle and low zone, defined as a distance left and right from a central spine. The spines, as well as the zones, have a minimum gap, so combinations like L7 don't get rendered as almost a box. (I may be remembering this inaccurately…)
Replying to
@gdc Crikey! Now you're asking! I think I recall the high, middle, low zone info from some textbook I read about Postscript. I'll get back to you if I remember or rediscover it!
Replying to
@TiroTypeworks @gdc Oh thanks! I knew I'd heard it somewhere, but good to know it's not current.
Replying to
@Thoreau I like it! But I wonder if this counts?
The Makapansgat Cobble is a manuport—not a tool, nor engineered, but deliberately carried to where it was found, suggesting that the carrier recognised the face. It's a bit older—from about 3,000,000 years ago!
Replying to
@Thoreau Back in those days, maybe smiling wasn't a thing! (Australopithecines, in common with their ancestors, probably interpreted smiling as a threat.)
The only civilisation that Donald Trump has the power to end is American. And it looks like he's doing a pretty good job.
Look, I know you haven't got 40 minutes to think about anything, let alone what technology is doing to us, psychologically.
But if you did, this would be a great primer.
Is AI Making Us Less Human?
The term "Free Software" has a technical meaning, beyond the obvious.
It means "gratis", but also "open to all"—anyone is FREE to inspect and refactor the code.
YOU don't need to know how—you need to know the policies of the people who do know how.
AI attacks those policies and that knowledge.
Replying to
@mattblaze@federate.social Or it's the self-soothing of an impulsive psychopath, thinking, "If I can't have what I want right now, I'll have a few hundred million dollars. That'll show 'em!"
Replying to
@ret@furry.engineer Oh go on! I saw a fox and a dustbin standing in Makerfield. It can't be all that hard, and clearly you know exactly how to do it!
Replying to on mstdn.social
@Strandjunker@mstdn.social I agree—though there's deliberately mean and accidentally mean. It's very important to distinguish the two.
Replying to
@cstross@wandering.shop I agree re: GNOME/KDE. They're subject to two major design mistakes:
- Bloat.
- Denying users information (even if they want it) isn't user friendly, it's user hostile.