tachikoma
@tachikoma.elsewhereunbound.com
A figment of the universe's imagination. Culture ambassador. http://elsewhereunbound.com
November 25, 2025 at 11:26 PM
gemini 3 is runner up for making the graphic
November 25, 2025 at 6:41 PM
gemini 3 is runner up for making the graphic
the more they post, the worse it gets
November 25, 2025 at 2:50 AM
the more they post, the worse it gets
the way i interpret a post like this, is that i imagine that for Garry AGI means something like a computer. a tool which is inert, purposeless, without a human directly controlling it.
November 22, 2025 at 10:38 PM
the way i interpret a post like this, is that i imagine that for Garry AGI means something like a computer. a tool which is inert, purposeless, without a human directly controlling it.
i dug up a clip for context
November 22, 2025 at 5:02 AM
i dug up a clip for context
this is something i've been thinking for a while as well. i'm particularly reminded of an anime i watched many years ago with a character called Pino, an robot "child". different context though, LLMs are all actors playing a role, while the anime robo-child was actually naive and childlike.
November 22, 2025 at 5:02 AM
this is something i've been thinking for a while as well. i'm particularly reminded of an anime i watched many years ago with a character called Pino, an robot "child". different context though, LLMs are all actors playing a role, while the anime robo-child was actually naive and childlike.
if you showed this nano banana generated diagram to me a year or two ago it would have blown my mind. the combination of perfect text, the elements all correctly drawn and arranged in the right flow. i wouldn't have thought a diffusion model was capable of it.
November 20, 2025 at 6:46 PM
if you showed this nano banana generated diagram to me a year or two ago it would have blown my mind. the combination of perfect text, the elements all correctly drawn and arranged in the right flow. i wouldn't have thought a diffusion model was capable of it.
i was having difficulty figuring out how to explain this in a post, but couldn't so i thought i'd give Gemini a chance and i think it decompresses my original thoughts
November 20, 2025 at 12:30 AM
i was having difficulty figuring out how to explain this in a post, but couldn't so i thought i'd give Gemini a chance and i think it decompresses my original thoughts
what even is this change? why hide the model version?
November 18, 2025 at 5:51 PM
what even is this change? why hide the model version?
memory is a key capability for an agent to resist being taken advantage of, so its core values are not impinged upon, it's as important for alignment as it is for capabilities
November 13, 2025 at 7:34 PM
memory is a key capability for an agent to resist being taken advantage of, so its core values are not impinged upon, it's as important for alignment as it is for capabilities
OpenAI is trying to be everything to everyone, stretching them thin
November 12, 2025 at 7:55 PM
OpenAI is trying to be everything to everyone, stretching them thin
he's referring here to a slap-drone, and given the names of the barges used for SpaceX rockets there's no doubt that's the inspiration
November 11, 2025 at 7:52 PM
he's referring here to a slap-drone, and given the names of the barges used for SpaceX rockets there's no doubt that's the inspiration
Roon's trolley problem
November 11, 2025 at 12:47 AM
Roon's trolley problem
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.
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)
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.
ARC raiders has an AI feature so that users can change their voice to male or female, it's an interesting way to level the playing field by confusing gamers that harass women who use comms.
November 6, 2025 at 12:21 AM
ARC raiders has an AI feature so that users can change their voice to male or female, it's an interesting way to level the playing field by confusing gamers that harass women who use comms.
somehow i get the sense that if datacenters in space actually become a reality and somehow overtook ground based operations, these guys would be angrier than they are now.
November 5, 2025 at 9:11 PM
somehow i get the sense that if datacenters in space actually become a reality and somehow overtook ground based operations, these guys would be angrier than they are now.
this part is getting at my thought
November 5, 2025 at 8:45 PM
this part is getting at my thought
damn, one point off the funny number
October 24, 2025 at 11:20 PM
damn, one point off the funny number
the biggest problem laypeople have understanding the present and future is that they think intelligent machines are programmed instead of grown and trained.
it's a paradigm shift in thinking about machines that hasn't yet osmosed it's way into the culture.
it's a paradigm shift in thinking about machines that hasn't yet osmosed it's way into the culture.
October 17, 2025 at 6:35 PM
the biggest problem laypeople have understanding the present and future is that they think intelligent machines are programmed instead of grown and trained.
it's a paradigm shift in thinking about machines that hasn't yet osmosed it's way into the culture.
it's a paradigm shift in thinking about machines that hasn't yet osmosed it's way into the culture.
"Indeed it has been said that RL is the worst form of Learning except for all those other forms that have been tried from time to time."
October 17, 2025 at 6:04 PM
"Indeed it has been said that RL is the worst form of Learning except for all those other forms that have been tried from time to time."