∿

Yo brother, legal team confirmed that we can't work with minors rn

I don't think this will happen, but a truly poetic way for bluesky to die would be for people to get its devs fired one by one until the site implodes.

oh AI is rapidly changing the scope of your job? that’s cool. that should make you cooler

why would I need to defend AI when it will just continue to make me hundreds of billions of dollars anyway

no

just open an issue

github.com/bluesky-soci...

you can ask claude code to use a Redis instance to store realtime traces of what it does and allow for more granular control — and an interface (this one uses tauri) to engage with it on a more granular level. and with git worktrees you can spawn as many simultaneous Claude Codes as you want

AI is just autocomplete. It will never be smarter than humans. Also Eliezer Yudkowsky was cancelled for jerking it in a Sports Authority flagship store, please do better

Fired a gun into the dark, but of course that’s what it is

You do not have to listen to Cartoon Americans about what LLMs can and can’t do. You do not have to perceive babyfurs at all. You can erase these types of people from your mind.

just checking in, what happened now?

its crazy sometimes on here I’ll have guys with a laundry list of depression / other mental illness in their bio DM me with some anti-tech or anti-AI stuff and its like, I’m literally feral and you’re walking into the PvP zone with less than 5% hp. what are you doing

Now let me design the plan with a Plan agent. [...] Excellent analysis from the Plan agent.

claude giving claude a medal never gets old

We really gotta get locked accounts and group DMs on here ASAP, it’s not healthy for the Discourse that everyone on here can act like this in public and anybody can screenshot.

huggingface.co/papers/2602....

I said “hardly”

Wow this website really seems to be growing in the amount of normal AI people who are here

I don’t think people understand that there are a few examples of Codex 5.3 and Claude Opus 4.7 producing at-human results in superhuman time. I think even fewer people want to accept this is happening

He sways public opinion and therefore matters. Wishcasting doesn’t work

I’ve hardly seen someone in public so committed to consciously lying for financial gain

The worst thing about Ed Zitron is that he’s objectively a bad person

A single kernel can be 200,000+ LoC.

He’s wrong. We have Claudes writing CUDA kernels unsupervised for days at a time that have anywhere from 80% to 240%+ speedup on certain hot paths.

software is clearly not over — it’s just entering its luteal phase

short political satire film by Claude “about ICE”

this is real work because its cool

I gave Claude Opus 4.6 access to Replicate and ElevenLabs in its entirety and every hour it makes a video about "whatever".

This is the first one it made, ocean.mp4

“The Floor is Lava” but with all organic material

Homie Vulgaris Bonus: Its all connected YouTube video by HomieVulgaris

Had a blast recording with @birdrespecter.bsky.social following the Mayo, folks where it leads is not good. Sub to the homie.gives for the full elsiode!

There's plenty of data that waymo is safer than average US drivers

But even ignoring that, just think—waymo does 500k trips/week now, if they had were even close to the crash rate of normal drivers they'd be killing a pedestrian like every other week
www.understandingai.org/p/human-driv...