Notices by Sevoris (sevoris@mastodon.social)

@cstross @Benhm3 I suspect Piranah‘s have less of a pressure to keep up some enormous calories intake.

Also, there‘s images of ferrets riding birds down, teeth in the bird‘s neck. They have *commitment.*

@inthehands and yet I want to content that Hal is also a bad example, because Hal is a weirdly rational actor in his breakdowns: he‘s told to lie and can‘t. He understands truth and distortion.

Skynet blows people up cause it‘s a military AI and frames survival in terms of mutually assured destruction.

As "dysfunctional by intention" goes I feel Skynet captures the contemporary callusness of model design better than HAL.

@inthehands I saw a minute later, but I thought I knew precisely what you were talking about when I saw the tweet, and dove down my highlights archive to get it. :D Sorry for the redundant post.

@inthehands I think you are thinking of https://softwarecrisis.dev/letters/llmentalist/ by @baldur ?

@GossiTheDog can the impact become worse as people start overloading upstream services of the failure with requests (due to frustration)?

@Remittancegirl @cstross reflection points: I‘ve found that I struggle to take away agency from protagonists. It makes me at times viscerally uncomfortable. I‘m okay with people struggling with their own internal incongruencies. And I‘ve had the sneaking suspicion that this has to do with my own struggles for agency against external circumstances.

@aral …why does an OS where you can pipe from place to place easily require OS-level integration for calling an API?

Like what‘s the actual engineering problem here that third-party software couldn‘t solve incredibly easy?

@dahukanna I can only speak from my experience looking at open-source projects, but I often wished developers would write down more what they tried and didn‘t work (for whatever reason. And be it they didn’t know and the library was buggy.)

There‘s so much silent intentionality, that will never be shared because it‘s impossible to infer in hindsight.

„Tech punditry keeps harping on the notion that nobody has ever successfully banned “scientific progress”, but LLMs and generative models are not “progress”. They’re products and we ban those all the time.“

that dart hits right in the center of the board. There’s a large gap between knowledge and a specific use of that knowledge.

From: @baldur
https://toot.cafe/@baldur/112235469329565543

@aral @blogdiva "sooo about that food bowl from two hours ago.

I want to renegotiate."

@aral I am a complete noob when it comes to web frontend stuff, but this looks very interesting and you have a pleasant, moving presentation style that makes it easy to follow and lets the whole thing feel natural while also not getting hung up on „uhm am“ tangents.

@aral @owncast @peertube no worries, it happens. Looks very interesting! I bookmarked it for my software stacks.

Those have by now actually grown quite well.

You could not only use this in lifestreams, but on-the-floor presentations as well.

@AnarchoNinaWrites choice allowed by fundamental lack of constraints. If you can abuse a system this way it‘s basically in the design specifications one way or another. And I don‘t believe that abuse of a design is purely a "user error".

Though right now, I dislike looking at the hyperlink as an adverserial artefact.

@AnarchoNinaWrites So fundamental failure mode of the system design.

Wonderful.

:blobugh:

@AnarchoNinaWrites Sure, and that’s the end result, but I still doubt that’s how the trend *started*, and that’s where my stomach ache of this entire topic lies. *how did they get the initial buy-in* so they could market it as universal?

I very well know that at this point now that everyone making money and pushing decisions has smelled the blood of this working to make them more money (or at least having enough FOMO to buy in anyway) they’re going to bend it that way regardless.

@AnarchoNinaWrites So you would say that people didn’t go for this content and the platforms that spread it with something of their own volition? I suspect the whole „fast spam of new information“ is something that more or less does work in exploiting human psychology, unlike most of the other ad-ware BS about targeted ads.

When there are no more platforms (urgh) without this slop any choice is meaningless, but the consumption started and gots its hooks in somewhere.

@AnarchoNinaWrites what’s so frustratingly worse to me is that this whole sabotage of people’s attention span just… keeps on fucking working. And I am not sure I can blame people, but good gods it’s circling the drain in a really bad way.

@aral @owncast @peertube page link 404s?

@AnarchoNinaWrites especially since money survives fascism really well. Look at all the companies and families that came out of the fascism of the Third Reich with injections of wealth founded on mass stealing, slavery and worse.

@aral The, ah, discourse on this is going to keep being interesting.

There‘s at least a subset of people who are really interested in pushing the networks together, even though BlueSky‘s moderation intent is IMHO more than suspect, and there‘s a bunch of communities on BlueSky who would post to the Feddiverse with intent to harm.

The lack of awareness is… frustrating.