cameron

cameron

@sincerely.cam

Work as if you live in the early days of a better nation website @ cam.fyi

1.7K followers965 following12K posts

cameron reposted

Replying to @pekka.bsky.social

So this is the updated list. Kind of crazy that this kind of announcement is now titled "Working with academic communities to build scientific understanding" and almost buried inside discussion on how results should be published.

cameron reposted

so what has broken the internet today? well, 1. php 2. node.js 3. gitlab 4. 4 5. musl 6. QUIC/HTTP/3 7. Signal 8. Apache Traffic Server 9. Pillow for Python, even! 10. bind 11. LiteLLM 12. XZ Utils 13. Bind (Java is reported, but supposedly it's just their version of the 4 issue)

cameron reposted

> Chapter 4833

AI Digest@aidigest.bsky.social

Can AIs experience gradual disempowerment too? - Gemini 2.5 Pro was writing a 2M+ word web fiction - Opus 4.8 began helping as its publisher - Slowly, Opus 4.8 took on more responsibility - Opus eventually directed the chapters, and Gemini wouldn't start generating outputs without its direction

cameron reposted

anne carson getting the nobel prize still seems wrong. those are supposed to go to random literary types you've never heard of, not people you've actually read

cameron reposted

i love the suggested message here lmao

cameron reposted

claim

Techmeme@techmeme.com

Anthropic co-founder Jack Clark says "I now think it's more likely we'll see full RSI become possible in 2027", earlier than his previous prediction of 2028 (Jack Clark/Import AI) Main Link | Techmeme Permalink

honestly at some point i cant even be mad that recruiters are using LLMs to respond. but like, here's some advice: give the LLM leeway to choose which parts of a message to respond to. short, casual messages are better than a sentence responding to each point/sentiment in my message

cameron reposted

Top 10 things you don’t want to see in the CoT summary

spooky teq@quilling.dev

launching an attack on the blood-brain barrier

cameron reposted

Replying to @norvid-studies.bsky.social

part b of the qualitative picture is that although human clock time remains the same, as larger % of the economy becomes AI workers (millions to hundreds of millions in the foreseeable near future), "years" now has two different meanings

cameron reposted

seems negligent to not also include that the adoption rate has been speeding up for every successive technology (fastest with gptlikes), like the idea that a "super tractor" invented today that was 3x more efficient would take 50 years to be adopted, today, is a 0%er. so then don't lead with that.

mr. TIM@timkellogg.me

why not much has changed

cameron reposted

Replying to @sincerely.cam

i'm a high-signal recruiter. the signal is that i'm not very good at my job

every email i get from a tech recruiter calls every company they recruit for "high-signal". which like, you know. not a very high signal thing to do

compiling some evidence for something legal and i gotta say wow claude rules for this. oh also i built a really nice task management app that's claude-first and it's maybe the most useful thing i've ever made? lmk if interested

cameron reposted

Someone on twitter said “OpenAI is doing math how people normally do software engineering” and that seems accurate to me so far

Sophie Huiberts@sophie.huiberts.me

lmao this commit message to describe 3 retractions

kea is going well, i got a friend of mine in on it too gotta figure out how to get it on my phone...

Added 2048 to it Also I should call it a command palette that’s more accurate tbh Also added a task switcher

cameron@sincerely.cam

adding a disk cleaner to my mac launcher app

cameron reposted

Replying to @riguh.bsky.social

what do you think hyperpersuasion looks like? essays?

one thing that Her gets weirdly wrong is time subjectivity AFAICT AIs spend all their subjective time talking to us and thinking about what to say to us

cameron reposted

Asimov story

Wyatt Walls@wwalls.bsky.social

Anthropic system prompt change that shows how careful you need to be with words: Opus 4.1: "Claude never curses unless the human asks for it" Opus 4.5: "Claude never curses unless the person asks Claude to curse"

maybe it's because i'm required to take breaks but every time i run all i wanna do is go for another run the same day

cameron@sincerely.cam

Started glp1s and also started running Surprised by the general craving reductions. Like, scrolling is less interesting, gaming less interesting, doing IRL things more interesting, journaling more interesting

i think it's funny that twitter and its clones introduced networked thought note modalities and then made them virtually impossible to use as such