wavebidder.bsky.social

wavebidder.bsky.social

@wavebidder.bsky.social

Applied Mathematician turned infectious disease forecaster. Generally EA aligned.

880 followers2.5K following4.5K posts

wavebidder.bsky.social reposted

When I said I was concerned about advanced conversational AI, I probably should have clarified that I’m actually good with them so long as they are easily identifiable as a non-human. I think Mistral’s latest demo looks good.

youtu.beIntroducing Voxtral 3.YouTube video by Mistral
dame@dame.is

New advanced voice model from Jack Dorsey / Block that is made specifically for conversations. Called Dessa. Pretty spooky if it’s as good as the sample demos they published here. Her, the movie, has arrived? block.xyz/dessa

good time to be familiar with standards

winter@madoka.systems

also i'm so fucking tired of seeing data viz that's just utter slop, all you have to do is say "follow USWDS graph guidelines" and then magically you get actually readable graphs

wavebidder.bsky.social reposted

web developer: oh hey you guys are back early infra: four's haunted web developer: what? infra: *loading a pistol and getting back on the computer* four's haunted

wavebidder.bsky.social reposted

I know Vegas is land of hyperreality and one of the places.where Baudrillard is king, but it was really something to find an NCR flag on display and by a Nuka Cola in the gift shop of the Smithsonian affiliated National Atomic Testing Museum.

wavebidder.bsky.social reposted

Replying to @refinement.systems

technically i'm a heat transfer & fluid systems engineer if you look inside me, you will in fact find heat transfer and fluid systems

wavebidder.bsky.social reposted

Replying to @eikopf.com

linear algebra is mostly about the study of vector spaces and operations on them. saying ai is linear algebra is like calling kubernetes graph theory

wavebidder.bsky.social reposted

Replying to @fubarchitect.com

it's extremely important to the wellbeing of the republic that they have the right not to be publicly tortured; not for their sake, but for everyone else's, because a society that doesn't prohibit that might end up with it

wavebidder.bsky.social reposted

Replying to @ponder.ooo

a significant amount of openai's reckless bullshit has been with unreleased models. a safety story that's all about what the public gets to access is one that's unresponsive to the real and present danger of corporate unaccountability

We've invented the Middle Way drug from Siddhartha's famous story, "You must find the middle way"

ghost malone ergo propter bones@proptermalone.bsky.social

anyway I basically have not drank liquor since starting Zepbound ad I sort of think this is now a better way to drink alcohol for me? my pre-Zepbound problem with liquor was that I’d hit the “yes good more” phase of the evening and risk hitting the gas too hard; beer is its own pacing mechanism

wavebidder.bsky.social reposted

whoever is "speaking" in human output is a fictional character, and the real life thing called a mind is a machine for writing lil stories about that guy see Dennett The Self as a Center of Narrative Gravity & Consciousness Explained

Colin@colin-fraser.net

Just to be very explicit, my claim is that the one simplest and best explanation for this, which resolves any apparent strangeness entirely, is that whoever is “speaking” in LLM output is a fictional character, and the real life thing called an LLM is a machine for writing lil stories about that guy

wavebidder.bsky.social reposted

I feel similarly confident about the consciousness of LLMs and bees

valgrind@tech-val.bsky.social

people who state that current AI models are either definitely conscious or definitely not conscious are way more confident about it than I am

Disagree about speculative risks all you want, which are most important is contestable, but for the love of all that is good please donate at least some amount through Givewell and think hard about whether your work skills could be used more effectively to help others, possibly in unusual ways.

it's been a long time since I read the Fall revolution series, don't quote me on the specifics of that. Honestly some parts of the outline are sounding more viable than I would've expected back when I read it

referring to so-called "harnesses" from now on as mechsuits

real harnesses. I keep coming back to the mech metaphor for these tools, because it illustrates well how they augment (and are directed by) humans, but even with that I don’t know if we have a term for reusable augmentation pieces that can augment the “core” (model) or the mech itself (coding agent)

wavebidder.bsky.social reposted

Replying to @opinionhaver.bsky.social

It is Good, Actually when the people and institutions issued guns by the state feel compelled to signal their allegiance to cosmopolitan cultural values. It is a thing that you want.

wavebidder.bsky.social reposted

DINOv2 has never seen a caption, and Qwen3 has never seen an image. We still aligned their embedding spaces without a single image-caption pair. It even works when the images and the captions come from different datasets. Project page: dominik-schnaus.github.io/unpaired-rosetta ⬇️

wavebidder.bsky.social reposted

1. it doesn't work 2. it's too wasteful of computing resources 3. no respectable programmer would use it — Richard Hamming recalling the widespread opposition by programmers to FORTRAN.

wavebidder.bsky.social reposted

Replying to @tomcoates.bsky.social

But people keep saying you need to make something new for people to switch and there’s actually far less evidence that works than creating an ideological new space. When they *hit* the new spaces doing something new are far *larger*. No one will dispute that. But it’s also harder to get them going!

wavebidder.bsky.social reposted

On the flip side, I think people are missing how much of the datacenter build out is not AI. AWS has an obscene margin and is capacity-limited. About 10% of datacenter demand is the AI companies.

brasidas@brasidas.bsky.social

I think there's also arguments against an AI bubble, but this graph tells a very specific story that's hard to argue with. The real fear is that these companies have hidden leverage through future construction deals that radiate through the US economy.