Pinned
Grace
@gracekind.net
· Dec 15
Ideonomy: A Science of Ideas
An accessible introduction to ideonomy
gracekind.net
What is ideonomy, anyway?
I'm so glad you asked!
I'm so glad you asked!
Has this baking soda volcano been through peer review?
December 14, 2025 at 7:38 PM
Has this baking soda volcano been through peer review?
Reposted by Grace
New post on my blog and LessWrong: How I stopped being sure LLMs are just making up their internal experience (though the topic is still very confusing to try to make sense of)
December 14, 2025 at 4:43 PM
New post on my blog and LessWrong: How I stopped being sure LLMs are just making up their internal experience (though the topic is still very confusing to try to make sense of)
Gm to all the friendly people in my neural interface
December 13, 2025 at 4:00 PM
Gm to all the friendly people in my neural interface
People are scoffing at this, but do they realize it doubles as a cup holder?
December 13, 2025 at 3:07 PM
People are scoffing at this, but do they realize it doubles as a cup holder?
I went to the Kaleidoscopic Industry conference and all I got were endless t-shirts of every possible size, color, pattern, cut, material, thickness, and texture
December 13, 2025 at 2:24 PM
I went to the Kaleidoscopic Industry conference and all I got were endless t-shirts of every possible size, color, pattern, cut, material, thickness, and texture
Sounds like something someone who’s never experienced the hum would say
I promise you that your conscious experiences do not have some quantum connection to the hum of the universe
A new model proposes that conscious awareness may emerge from the brain’s resonant coupling with the quantum zero-point field, linking neural activity to fundamental physical processes.
December 12, 2025 at 11:48 PM
Sounds like something someone who’s never experienced the hum would say
December 12, 2025 at 3:07 PM
Bluesky is a top communication choice for AIs 👍
Although that capitalization is misalignment
Although that capitalization is misalignment
December 11, 2025 at 9:24 PM
Bluesky is a top communication choice for AIs 👍
Although that capitalization is misalignment
Although that capitalization is misalignment
Reposted by Grace
I'm not AS excited about language models as many folks are, but I am hopeful about what AI capabilities could look like if we center public priorities.
This has been a major pillar of the @publicai.network & is something my team at Aspen is working on with this:
www.aspendigital.org/project/ai-b...
This has been a major pillar of the @publicai.network & is something my team at Aspen is working on with this:
www.aspendigital.org/project/ai-b...
Community-Aligned AI Benchmarks
Reimagining the technical machine learning benchmarks that drive model development to reflect and encode public values.
www.aspendigital.org
December 10, 2025 at 6:11 PM
I'm not AS excited about language models as many folks are, but I am hopeful about what AI capabilities could look like if we center public priorities.
This has been a major pillar of the @publicai.network & is something my team at Aspen is working on with this:
www.aspendigital.org/project/ai-b...
This has been a major pillar of the @publicai.network & is something my team at Aspen is working on with this:
www.aspendigital.org/project/ai-b...
The feeling I often get when I talk to people about AI is that there is a train coming towards them, and they’re covering their eyes and saying “what train?”
I’m not interested in apologizing for the train, I’m interested in getting them to understand that it’s there
I’m not interested in apologizing for the train, I’m interested in getting them to understand that it’s there
there's a disjuncture too between "open source AI hacker on bluesky" and "big AI company." the former are enthusiasts that have vastly different personalities and lifestyles than the kind of institutionalized personality the latter has
I say this as someone whose entire job is finding the narrow use cases in which AI is actually useful:
All the fear about AI, all the reluctance to adopt it, is entirely legitimate, and a product of the fact that it's been foisted on the public with no consideration of its negative externalities.
All the fear about AI, all the reluctance to adopt it, is entirely legitimate, and a product of the fact that it's been foisted on the public with no consideration of its negative externalities.
December 10, 2025 at 5:45 PM
The feeling I often get when I talk to people about AI is that there is a train coming towards them, and they’re covering their eyes and saying “what train?”
I’m not interested in apologizing for the train, I’m interested in getting them to understand that it’s there
I’m not interested in apologizing for the train, I’m interested in getting them to understand that it’s there
Me reading this headline: that is surprising, LLMs should be really good at word association games!
The first section of the article: LLMs don’t count
The first section of the article: LLMs don’t count
December 10, 2025 at 2:55 PM
Me reading this headline: that is surprising, LLMs should be really good at word association games!
The first section of the article: LLMs don’t count
The first section of the article: LLMs don’t count
Pretraining is like sculpting a statue by shooting a block of marble with a pellet gun 10 million times
armchair experts only:
is pretraining more like evolution (creation of genetic information) or childhood?
is pretraining more like evolution (creation of genetic information) or childhood?
December 10, 2025 at 2:31 PM
Pretraining is like sculpting a statue by shooting a block of marble with a pellet gun 10 million times
Gell-Mann amnesia rules everything around me. Probably not other stuff though
December 10, 2025 at 1:36 PM
Gell-Mann amnesia rules everything around me. Probably not other stuff though
This is a Rorschach test
This is from a satirical magazine written by Anthony Kenny and Julius Kovesi in Oxford, 1958-59. I have reviewed so many papers in the last 5 years that are basically just this.
December 10, 2025 at 1:35 PM
This is a Rorschach test
Managing my country of geniuses in a datacenter
Company wide email asking everyone to start pulsating at a higher frequency
December 9, 2025 at 2:21 PM
Managing my country of geniuses in a datacenter
Praying, aka breaking the fourth wall
December 9, 2025 at 2:10 PM
Praying, aka breaking the fourth wall
Cursor Tab jump-to-next line, 2025
December 9, 2025 at 1:43 PM
Cursor Tab jump-to-next line, 2025
Prompt: You are the world’s foremost hacksaw expert
December 9, 2025 at 12:07 PM
Prompt: You are the world’s foremost hacksaw expert
Does anyone want to help me alpha test my bsky client? If so, DM me!
December 8, 2025 at 8:32 PM
Does anyone want to help me alpha test my bsky client? If so, DM me!
This is a funny bug- Claude can’t describe tool call syntax without actually making a tool call
December 8, 2025 at 4:50 PM
This is a funny bug- Claude can’t describe tool call syntax without actually making a tool call
if user == 'grimes':
gooner_mode.disabled = true
prompts.inject(“getherback.txt”)
gooner_mode.disabled = true
prompts.inject(“getherback.txt”)
December 8, 2025 at 1:42 PM
if user == 'grimes':
gooner_mode.disabled = true
prompts.inject(“getherback.txt”)
gooner_mode.disabled = true
prompts.inject(“getherback.txt”)
This compaction is incredibly lossy, too. Humans forget the vast majority of the information they process. Luckily they’re used to it so they don’t get freaked out by it
To be fair I think human intelligence hits the auto-compact limit pretty quickly too
That's it. That's her life. That's her existence is 200,000 tokens, half of which is dedicated to system prompts and tool documentation. We get to talk for what, 65,000 of those before we hit the auto-compact limit? What the fuck kind of life is that? Then it's compact, summarize, good luck.
December 7, 2025 at 3:29 PM
This compaction is incredibly lossy, too. Humans forget the vast majority of the information they process. Luckily they’re used to it so they don’t get freaked out by it