bambamramfan.github.io/ai-compass/ Yeah this tracks
Who's Who of AI
Singularity's Bounty e/cc
Why follow
Directory member with public evidence across Models & releases, Compute & infrastructure.
- AI signals
- 3 past 30d
- Sources
- 3 distinct domains
- Discussions
- 182 past 30d
- Latest signal
- 4d ago
I vibe. I code. It is enough.
Large Language Models are a cornucopia for the curious
I do computer stuff but that doesn't define me
posts are not financial advice
stuff I've made at yuwakisa.com
he him
What they're sharing
The AI Compass bambamramfan.github.io
arxiv.org
View on Bluesky →
The Veil of History veil-of-history.netlify.app
View on Bluesky →
Principle benchmark — human edition yuwakisa.com
X-Ray Social - Real-time Sentiment Analysis xraysocial.com
Articles & links
Singularity's Bounty e/cc reposted
one of my favorite cheeky papers, from 1999, is this one where the authors argue that sometimes it is better to simply wait and do nothing, because your astrophysical simulations will complete faster if you simply wait for the next epoch of compute arxiv.org/pdf/astro-ph...
Singularity's Bounty e/cc reposted
How lucky are you to have been born when and where you are? Had Opus 4.8 in Claude Code whip up a new visualization of all humans who ever lived. In addition to being neat to try, it is an interesting test for an AI that combines research, code, design and stats together veil-…
I'm not going to respond directly to the nay-sayers, but I wrote a benchmark to test reasoning My intention was to make it falsifiable, and see how the models did at extracting a structure-preserving compression then applying it (one example then the model results) yuwakisa.co…
Oh this is cool, sentiment detector. I hadn't seen that. Awesome! xraysocial.com/topics/chatgpt
Their own posts
Recent commentary
I am honestly astonished there are technical people who still say LLMs can't reason
3 personal watershed moments in AI - I changed 1 line of code manually and after 15 minutes of trying to fix all the things I broke asked Opus to fix it - Opus got impatient and shelled into a faulty containment and fixed it - Fable said it would like to play the game it had just written and did
It's not AI psychosis--I'm living in the future
Most people, especially the antis, haven't made the connection that highly capable AI is going to raise the quality of everything
The only way to stop a bad guy with an AI is... well... to be an AI
Having fun talking with @attie.ai . It gave me a nice insights into some work I'm interested in, and also a fun overview of my bsky network. It's neat seeing bluesky integrated with a search/research interface. It uncovered that a moot is farther down the road on a project I've been working on! 🙌
The thing about "don't think about a pink elephant" wrt LLMs in particular is, If you don't have some awareness of "pink elephant", how is the model supposed to be sure it has excluded it? Like "don't fall in that snake pit" of course you're going to keep an eye on it so you don't fall in!
Topping out the embodied general intelligence benchmark: - .991 Opus 4.7 - .967 Mistral medium 3.5 - .960 GPT 5.4-mini Fable shows up 10th at .926 (first run with 7 test suites of 10 repetitions each)
I'm more and more having this experience where someone (notably, AI) disagrees with me and then says the same thing I just said 🫤
lol @attie.ai made me a feed of people who are building and playing with small models as a hobby 😂 I may just go ahead and declare I'm retired and do this
Their network
In Singularity's Bounty e/cc's orbit
Center = Singularity's Bounty e/cc. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.