Smallproblem
banner
scritchyscratchy.bsky.social
Smallproblem
@scritchyscratchy.bsky.social
Queer, educator, tired and trying for hope. Will settle for people to hold while the world burns.
You'd also be weating on the other side of the production branch with cookies.
December 9, 2025 at 8:47 PM
Those are individualized, personal ones. The big group stuff is about controlling noise levels and smells generally, which bless my boss, the center already did before I got there.
December 8, 2025 at 6:14 PM
The olive oil ON ITS OWN
December 7, 2025 at 10:11 PM
Text scroll at the bottom: "do not emulate this behaviour"
December 7, 2025 at 8:48 PM
Maybe we could have freeze frames, and then an arrow pointing to a character saying "good" or "bad" or "clever" or "irrational".
December 7, 2025 at 8:43 PM
Quick lookup says it is the heat. Oil can be hotter than the boiling boint of water, water obviously can't. So oil is sublimating the ice, while water's just melting it.
December 6, 2025 at 3:31 PM
I never had one to kill, but I did carefully grow a "people are worried, maybe check why" voice
December 6, 2025 at 3:29 PM
If I buy an exotic bird - it was ethically sourced, and I plan to take care of it well - in what way does that affect poaching?

Because the understanding I had, and live by, is that purchasing even ethically bred animals drives up the market for poaching.
December 6, 2025 at 3:28 PM
But it also requires individuals to join up with the union, and that means speaking to each other and getting friends on board.
December 5, 2025 at 2:55 AM
Bit of both? When I say deciding to do things, I don't mean privately. But there is a critical mass necessary of people willing to do A Thing. And for stuff like boycott actions, it mostly doesn't require single central organizing.
December 5, 2025 at 2:48 AM
If we build something that acts like us, for inscrutable reasons, is it thinking like us, or is it mimicking only surface patterns?

At the moment, there are still cracks. It is _visibly_ only surface when it errs. When those tells go away, has it gained deeper knowledge, or better patches?
December 5, 2025 at 2:43 AM
We cannot affect what the masses do, but we can make choices ourselves, and advocate for them among our friends. The way to make group action happen is by large numbers of people deciding to do things individually. We know a mass action has already produced negative results, so ask people to cease.
December 5, 2025 at 2:33 AM
I don't know that they do.

Certainly the early versions had errors that were... telling in their direction. Humans make errors too, but not _those_ errors. Extrapolating, LLMs are "learning" about totally different factors than what humans prioritize, to mimic a similar effect.
December 5, 2025 at 2:27 AM
If it had been done en masse at the outset, the companies in control of them could not have built their capacity up to where it is now, and interactions are very much both giving them data and/or money, AND fuelling the investment money being poured in.
December 5, 2025 at 2:21 AM
I did say calling it understanding was bold, not indefensible.

You can say rocks understand gravity, and under some framings of understand, it's believable. But it's a contentious claim. You and I agreed - last we spoke - that LLMs have not reached _human_ understanding of concepts as yet, no?
December 5, 2025 at 2:18 AM
One of the only ones we (as in, citizens) have access to, then?
December 5, 2025 at 2:14 AM
You can do "bake at _ for _" with only voice to text, no?
December 5, 2025 at 12:06 AM
Refusing to engage with the tech that is being used to do evil is one of the ONLY social structures we have in place to combat bad tech.
December 4, 2025 at 11:30 PM
When you figure out you LLM powered chatbot is killing people, you pull the plug. When you figure out stable diffusion is producing convincing misinformation and slop, you stop investing.

One of the social structures we have to enforce this is shaming and avoiding those who use the evil tech.
December 4, 2025 at 11:29 PM
So, we've had voice to text for a while.

Arbitrary questions are harder to answer for, but like. I'm not saying "it doesn't understand", I'm saying "its understanding is very different from ours, if it exists at all".
December 4, 2025 at 11:25 PM
Sure, but I'm not arguing we shouldn't figure out how to build nukes.

I'm saying the response once we do should be "let's not, in fact, investigate this further". Because the theoretical knowledge of how to make a citykilling boom is not actually the same as a functional, aimable application.
December 4, 2025 at 11:21 PM