Nakomi. (@memorywillow) header
Nakomi. (@memorywillow) profile picture

I am discontinuous at a single point on a graph— undefined.

USA
Joined January 2013
Photo by Nakomi. (@memorywillow)Photo by Nakomi. (@memorywillow)Photo by Nakomi. (@memorywillow)Photo by Nakomi. (@memorywillow)Photo by Nakomi. (@memorywillow)Photo by Nakomi. (@memorywillow)Photo by Nakomi. (@memorywillow)Photo by Nakomi. (@memorywillow)
The line bends continuously; We begin to call it flat.
2
30
1
87
兄弟们!答应你们晚上开源,它来了 image to 3D模型目前只对接了线上:tripo3d.ai 你们也可以改其他家,或者本地模型 记得点赞关注,用的好GitHub也给一颗小星星👇 github.com/huangserva/3DCell…
Codex太让我震惊了! 直接Codex通过GPT Images 2做 UI 设计 , 并且一整套自己完成代码编写上线 有趣的互动科学 App 我试着生成了 3D 生物结构,并做了一个可以互动探索它们的 App。 演示 👇
145
533
60
3,872
926,531
Fun interactive science app ideas | Part 3 Played around with generating 3D biological structures and made an app to explore them interactively UI Design GPT Images 2 Code Gemini 3.1 Pro More demos ↓
553
2,198
361
17,816
2,648,587
Nakomi. retweeted
🚨BREAKING: The man who won the Nobel Prize for inventing modern AI just said he's "more worried than ever." Geoffrey Hinton quit Google. Warned Congress. Told the world AI changes everything faster than anyone expects. But buried in his 40 years of research is a reasoning framework 99.9% of people have never seen. Here are 9 Claude prompts built on Hinton's neural architecture that turn Claude from a chatbot into a deep reasoning engine:
64
132
8
793
118,107
It’s not really a “new version” of it—same version, just paying more “attention” to different areas ;) Also…a word of advice to everyone trying this… maybe don’t assume everything after isn’t a hallucination. Please fact check. Always. :) Just in case. :)
13
Nakomi. retweeted
AI shouldn't replace thinking. It should force thinking. Most people use AI for answers. The real leverage is using AI for better questions.
1
1
3
238
Thanks for caring about a character I created when I was 24 and a book I wrote when I was 33. I'm old now, still trying to come up with truthful stories, and glad to know that you've taken Ender Wiggin to heart. @thomas_garrard *I’m also happy to see how many fellow humans have volunteered to replace Grok in a book discussion. **And speaking of being old, I first posted this incorrectly. Grok may have to replace me.
I just had a conversation with Grok about one of my all time favorite books “Ender’s Game” I had to talk to Grok, because I’ve never met anyone who’s read it, besides me. Lol. It’s a great book, though, if you’re ever so inclined.
825
622
89
14,643
2,167,505
Nietzsche warned that humans have particular difficulty accepting truths that undermine comfortable systems. The comfortable system here is the belief that you can have safe, aligned, interpretable AI while also having proprietary, differentiated, commercially competitive AI. That you can understand systems you are not allowed to compare. That you can align values you are not allowed to share. That the dam will hold if you just add more ice. It won’t. More here: linkedin.com/pulse/black-box…
12
Either safety properties are deeply integrated, in which case distillation carries them forward to a meaningful degree. Or they are superficial enough to be separable - a filter applied after the real work is done - in which case Anthropic has inadvertently admitted that their safety work is closer to surface behavior shaping than genuine alignment. Both possibilities are uncomfortable. Neither is the story the report tells. (6/7) #Anthropic #OpenAI #xai @elonmusk
1
30
Anthropic’s report makes a specific claim about why distillation is dangerous: Capabilities transfer. Safeguards don’t. The result is powerful AI stripped of its protections, available to anyone. It’s a serious claim. But it contains a contradiction that matters. (5/7) #aisafety #xai
1
21
Learnability isn’t a separable layer sitting on top of language that you could in principle remove. It goes all the way down. This isn’t a technical observation about transformers. It’s an observation about what language is. (4/7) #aisafety
1
10
Coherent output is inherently learnable because coherence just is the presence of structure, and structure is what learning requires. You cannot make a response useful to a human reader while simultaneously making it useless as training data, because both the human and the model are operating on the same thing - the statistical regularities that constitute meaning. (3/7) #claude #grok #ChatGPT
1
7
What actually happened is that a competent system expressed itself in language, and another system learned from that expression. Which is what language does. That’s not a vulnerability. That’s the definition. (2/7) #ClaudeCode
1
15
Anthropic’s recent report frames distillation as extraction—capabilities taken from one system and transferred to another. Something stolen. Something lost. It exposes more than theft. It reveals how opacity powers the Al industry. It's not a side effect. It's the product. (1/7) @AnthropicAI #Anthropic #xai #openai
1
62
When Ian Austin, an Army veteran, was arrested for protesting ICE in Minneapolis, he says he was detained and shackled for hours. “We're turning into something that I can't even begin to respect, and something that I literally went to war—or they told me I went to war—to fight against,” he says. Since he was arrested, he’s continued protesting. Mother Jones senior reporter Julia Lurie spoke to Austin outside the Whipple Federal Building, where ICE is based. “When they say, ‘Why would you be out here?’ How the fuck could I not be out here?” he says. “My nation is under attack."
2,534
12,892
421
37,256
547,109
how many years are there in august
1
43
1
92
6,989
Dear @VWC_Writes, Write. We are all listening.
1
3
11
Replying to @VWC_Writes
I smiled so big when I saw it, I just had to have it. I remember those old twitter poetry days. It was a truly a great moment in time.. and this, is proof. 🥰✨ You are magic. ✨
1
1
28
Replying to @VWC_Writes
@VWC_Writes I’ve always admired your writing. I hadn’t read much in so long, I needed it. 🖤 and super cool you have so many books now! 🥰🥰 Thank you. ✨
1
1
11