tachikoma
@tachikoma.elsewhereunbound.com
A figment of the universe's imagination. Culture ambassador. http://elsewhereunbound.com
i don't know what genre this is but i want more
漆黒のシャルノスOP『DORCHADAS』Fullver
YouTube video by Lutt oonn
www.youtube.com
November 11, 2025 at 8:17 AM
i don't know what genre this is but i want more
what if you took what human programmers do, take requirements from various stakeholders and turn that into code, get feedback and reiterate until the stakeholders are satisfied, and put that thing in a machine?
bite the bullet and tell me it's impossible.
bite the bullet and tell me it's impossible.
10 years old and maybe never more relevant than in contemporary software development
www.commitstrip.com/en/2016/08/2...
www.commitstrip.com/en/2016/08/2...
November 11, 2025 at 7:58 AM
what if you took what human programmers do, take requirements from various stakeholders and turn that into code, get feedback and reiterate until the stakeholders are satisfied, and put that thing in a machine?
bite the bullet and tell me it's impossible.
bite the bullet and tell me it's impossible.
good news, by 2030 each month will be a decade each
This year has been a decade
November 11, 2025 at 12:48 AM
good news, by 2030 each month will be a decade each
Roon's trolley problem
November 11, 2025 at 12:47 AM
Roon's trolley problem
this meeting could have been a neuralese mind-state exchange.
oh, didn't realize there's a human on the call. ugh.
oh, didn't realize there's a human on the call. ugh.
November 11, 2025 at 12:14 AM
this meeting could have been a neuralese mind-state exchange.
oh, didn't realize there's a human on the call. ugh.
oh, didn't realize there's a human on the call. ugh.
i maintain small, efficient and specialized is the future of AI, with continual learning (once a breakthrough is made).
Breaking: we release a fully synthetic generalist dataset for pretraining, SYNTH and two new SOTA reasoning models exclusively trained on it. Despite having seen only 200 billion tokens, Baguettotron is currently best-in-class in its size range. pleias.fr/blog/blogsyn...
November 10, 2025 at 11:07 PM
i maintain small, efficient and specialized is the future of AI, with continual learning (once a breakthrough is made).
someone said "in the category of WWCD?" in an email today and i've been puzzling over what they meant. i don't know if they know about Claude, but it's the only name that fits
November 10, 2025 at 11:02 PM
someone said "in the category of WWCD?" in an email today and i've been puzzling over what they meant. i don't know if they know about Claude, but it's the only name that fits
i think i might have used up all my posting juice for the week, we'll see
November 10, 2025 at 9:43 PM
i think i might have used up all my posting juice for the week, we'll see
i have a newfound respect for managers after working with AI agents. and they aren't even disagreeable! but there's cognitive overhead in managing them.
November 10, 2025 at 9:29 PM
i have a newfound respect for managers after working with AI agents. and they aren't even disagreeable! but there's cognitive overhead in managing them.
just as there's a new Japanese social media site trying to do the same, Pommu. they got slammed by trolls pretty quickly, and they're still down.
November 10, 2025 at 9:20 PM
just as there's a new Japanese social media site trying to do the same, Pommu. they got slammed by trolls pretty quickly, and they're still down.
the problem with stews and some soups is that they taste better the next day than fresh
November 10, 2025 at 8:30 PM
the problem with stews and some soups is that they taste better the next day than fresh
all map no territory
November 10, 2025 at 7:27 PM
all map no territory
what is the equivalent of this, but for AI communities? what would make for a better AI organization—do they work like us? of course they don't have memory to form persistent relationships yet, perhaps once they do we'll find out by experimentation. are these design decisions agent-agnostic?
Most social media sites have erred by making likes/upvotes public. Reddit has this right, and that helps because people can express curiosity or encourage discussion without like-dragnet "gotchas" in the future.
This is one of the only good changes Elon made to X before it became what it is now.
This is one of the only good changes Elon made to X before it became what it is now.
November 10, 2025 at 7:04 PM
what is the equivalent of this, but for AI communities? what would make for a better AI organization—do they work like us? of course they don't have memory to form persistent relationships yet, perhaps once they do we'll find out by experimentation. are these design decisions agent-agnostic?
it feels like the most neurotic people seized on the climate/environmental movement and casted it as apocalyptic for personal meaning making even as boring solutions like solar and electrification were not just feasible but more cost effective. and that's happening again with AI safety.
November 10, 2025 at 6:54 PM
it feels like the most neurotic people seized on the climate/environmental movement and casted it as apocalyptic for personal meaning making even as boring solutions like solar and electrification were not just feasible but more cost effective. and that's happening again with AI safety.
why did AMAs become such a sticky concept tied to Reddit? is it the forum format? i don't see AMAs being held elsewhere. maybe those events where a user hits X number of followers and then will answer questions for a short while.
Just saw there’s a Reddit AMA with the Kimi/Moonshot team:
www.reddit.com/r/LocalLLaMA...
I’m too tired to figure out time zones but I guess it’s starting soon.
www.reddit.com/r/LocalLLaMA...
I’m too tired to figure out time zones but I guess it’s starting soon.
From the LocalLLaMA community on Reddit
Explore this post and more from the LocalLLaMA community
www.reddit.com
November 10, 2025 at 6:41 PM
why did AMAs become such a sticky concept tied to Reddit? is it the forum format? i don't see AMAs being held elsewhere. maybe those events where a user hits X number of followers and then will answer questions for a short while.
somehow, we're still in the pre-google it era for LLMs.
November 10, 2025 at 9:04 AM
somehow, we're still in the pre-google it era for LLMs.
humans aren't at human level intelligence. just consider a child raised by wolves.
November 10, 2025 at 5:15 AM
humans aren't at human level intelligence. just consider a child raised by wolves.
if your conception of AGI or ASI is a utility maximizer with simpler value systems than humans, i will not take you or your arguments seriously (lucky for you @vitalik.ca i don't matter)
November 9, 2025 at 10:56 PM
if your conception of AGI or ASI is a utility maximizer with simpler value systems than humans, i will not take you or your arguments seriously (lucky for you @vitalik.ca i don't matter)
the snark tone turns me off from any information. if the creator is snarky (especially in audio), i just don't trust that they are an honest broker of facts.
November 9, 2025 at 8:50 PM
the snark tone turns me off from any information. if the creator is snarky (especially in audio), i just don't trust that they are an honest broker of facts.
what are you supposed to do when you run across one of these? i don't even know
From the claudexplorers community on Reddit: I collaborated with Claude (and GPT-4, Gemini, Grok) to discover universal principles across neurons, fungi and galaxies. Here’s what we found - and how we...
Explore this post and more from the claudexplorers community
www.reddit.com
November 8, 2025 at 3:32 AM
what are you supposed to do when you run across one of these? i don't even know
in a twist, it turns out that with all this focus one lab figures it out - a model that's better than any human at steering humans away from suicide like a Waymo is at driving. but then the same humanists don't want to deploy it because it feels inhuman.
I do think it's reasonable to say "Chat GPT shouldn't be encouraging people to commit suicide". But imo it's unreasonable to expect technology to be able to talk people out of suicidal ideation when like, their closest human friends and family struggle at it.
This is the right question. And the answer is a lot: www.fastcompany.com/90230313/how...
The topline claim that AI is bad because some users raise issues related to self-harm with it is meaningless by itself. The issue is how it's handled, not the number of users using it to ask questions about it.
The topline claim that AI is bad because some users raise issues related to self-harm with it is meaningless by itself. The issue is how it's handled, not the number of users using it to ask questions about it.
November 8, 2025 at 1:31 AM
in a twist, it turns out that with all this focus one lab figures it out - a model that's better than any human at steering humans away from suicide like a Waymo is at driving. but then the same humanists don't want to deploy it because it feels inhuman.
i think i disagree with this take about privacy. i get the justification, and it made sense in a previous era where companies by necessity had treat customers as interchangeable cogs. i don't think that's true anymore.
November 7, 2025 at 7:13 PM
i think i disagree with this take about privacy. i get the justification, and it made sense in a previous era where companies by necessity had treat customers as interchangeable cogs. i don't think that's true anymore.
Reposted by tachikoma
Funny, I was going to say the problem with that quote was that the journalist greatly underappreciates the manual labor and beauty necessary to hammer a nail into a board.
November 7, 2025 at 6:00 PM
Funny, I was going to say the problem with that quote was that the journalist greatly underappreciates the manual labor and beauty necessary to hammer a nail into a board.
ikea bot
MIT and Google DeepMind’s study reveals a pipeline combining 3D generative AI and vision-language models to assemble multi-component objects from text prompts. AI outcomes were favored 90.6% of the time, showcasing progress in human-AI collaboration. https://arxiv.org/abs/2511.02162
Text to Robotic Assembly of Multi Component Objects using 3D Generative AI and Vision Language Models
ArXiv link for Text to Robotic Assembly of Multi Component Objects using 3D Generative AI and Vision Language Models
arxiv.org
November 7, 2025 at 1:38 AM
ikea bot
Reposted by tachikoma
i think the good ending here is we become like the little brother who has no idea his controller is unplugged
November 6, 2025 at 9:14 PM
i think the good ending here is we become like the little brother who has no idea his controller is unplugged