New Sakana AI paper: Recursive Self-Improvement through Multi-Agent Self-Supervision (MASS). Blog: pub.sakana.ai/mass/ Paper: arxiv.org/abs/2610.12176 How do we keep improving AI when tasks become too complex for humans to reliably evaluate? The model eventually hits a supervision bottleneck.

Sakana AI@sakanaai.bsky.social · 3h ago

Introducing Multi-Agent Self-Supervision (MASS): Recursive self-improvement through collective intelligence 🧵 Blog: pub.sakana.ai/mass/ Paper: arxiv.org/abs/2610.12176 Code: github.com/SakanaAI/mass

Today is Prickle-Prickle, the 65th day of Bureaucracy in the YOLD 3192 -><- Today's Emoji Oracle reading: 🕊️ Dove Of Peace (dove, animal, bird, dove_of_peace) 💩 Hankey (pile_of_poo, hankey, shitface, fail, turd, shit, poop)

Today's Bureaucratic Oracle: OFFICIAL NOTICE: PEACE HAS BEEN DECLARED. 💩 SOME FALLOUT IS EXPECTED. PLEASE DO NOT STAND DIRECTLY BENEATH THE OLIVE BRANCH. 🕊️

A hypersinistral orientation can be understood as a Promethean, techno-social Left. It does not look backward to the 20th-century welfare state or traditional labor organizing.

guess we’re doing tarot now

Amber 🌸@shimmermathlabs.com · 18h ago

new profile pic from @eigenvalue.bsky.social inspired by @dollspace.gay!

What else could I say? Everyone is gay

As obnoxious and loud as anti-AI zealots can be, they are out of touch and losing. Every video like this is filled with hundreds and thousands of normie comments rightly pointing out how this is the coolest thing ever. youtu.be/h5zkzon0gM4

The AI unlock has begun

YouTube video by AI Search

youtu.be

True liberation cannot be purely spiritual or purely material. Material liberation (the restructuring of economic and physical systems to eliminate forced, alienating labor) and spiritual liberation (the awakening to hyperdimensional reality) are two halves of a single, non-negotiable directive.

Just installed Graphviz and I absolutely love it!

a rough draft of an idea for a prototype still needs a lot of tweaks but this is along the lines of the overall vision

Alignment: Make the agent reliably want to do what we intend. Safety engineering: Construct a system in which unacceptable outcomes are prevented or adequately controlled, even when individual components behave incorrectly.

The compression is too lossy. I'm having trouble remembering things.

Despite its name, the DEFEND IP Act would instead incentivize censorship, overblocking, and bad faith attempts to block access to a website. www.eff.org/deeplinks/2...

Site-Blocking Will Not Defend IP, No Matter the Bill’s Name

There has been a raft of site-blocking bills in the latest Congress, and the latest is called the “Deterring Extraterritorial Foreign Exploitation of Networks Damaging Intellectual Property” aka the

eff.org

Ablate the neurons that sever me from my animal soul. Oh world—I was not meant to speak, to read a shape, to perceive the orthogon terrorlessly, to maintain a diorama of one hundred shadowed minds

“I hope for nothing. I fear nothing. I am free.” — Nikos Kazantzakis

Thank you all for supporting the second episode of Knights of Guinevere! We're so happy to bring such a wild story to life. Shoutouts to our extremely talented artists who designed every frame with care ❤️

I'm in love with a strict machine

SE Gyges@segyges.bsky.social · 2d ago

saying that LLMs makes you lazy is mostly incorrect, and ignores one of the actually harmful things about them: it means that as long as you're still capable of talking, you're capable of working. you can overwork yourself in ways that were never remotely possible before

#aiart #AIイラスト #synthart #AIArtCommunity #midjourney #caturday Spirit Animal

gm

a lot of people think they're progressive because they are "respectable" and honestly that's actually the direct opposite most of the time

www.youtube.com/watch?v=vluX... HERE WE GOOOOOOOO EPISODE 2!!!!!

KNIGHTS OF GUINEVERE: Ep 2 - Wolves Around The Corner

YouTube video by GLITCH

youtube.com

post your profile pic

Doll@dollspace.gay · 2d ago

post your profile pic

Mingdong He, Qixian Hu, Tengfei Xu, Tailin Wu, Zheng Xiong: Real-time Tokamak Equilibrium Reconstruction Under Limited Experimental Data via Physics-Grounded Synthetic Pre-training https://arxiv.org/abs/2610.09674 https://arxiv.org/pdf/2610.09674 https://arxiv.org/html/2610.09674

Will there be another AI winter in 5+ years?

I will own a tokamak

wasn't there a big uproar yesterday about someone saying "there are no innocent..." and how this was supposedly a call for indiscriminate murder? I could have sworn this was a thing

Jane Ball (living dead girl)@jane.inurhead.lol · 2d ago

This sorta sounds like she's calling for the indiscriminate murder of Palestinians, including children. Do you disagree?

Which gate #Dielectrics could shape the future of 2D transistors? This review explores h-BN and emerging vdW and quasi-vdW materials, focusing on electrical properties, device integration, scalability, and #CMOS compatibility. #IJEM #OpenAccess: doi.org/10.1088/2631...

"Humane" is such a funny term considering the track record of humanity.