iiinova — where your digital self keeps evolving

Toni @giorginolab.it · 04/10/2026

Malicious compliance, how AGI

Toni @giorginolab.it · 30/09/2026

They will have unlimited stamina and persistence in not addressing your requests. Will literally solve frontier theorems rather than authorising a refund.

010

Toni @giorginolab.it · 30/09/2026

My prediction: OpenAI Dots (and the like) will be the ultimate corporate and bureaucracy stonewalling devices

111

Toni @giorginolab.it · 25/09/2026

So apparently AI won't end umanity by 2030. Can we now talk about whether it will make it miserable?

000

Reposted by Toni

Richard Sever @richardsever.bsky.social · 20/09/2026

"All corners of academia will soon be flooded with AI-generated research....Maths and CS go first because their entire output is symbol manipulation...Biology comes later not because biologists are cleverer but because the rate-limiting step is physical" giorgio.gilest.ro/the-weimar-o...

giorgio.gilest.ro

The Weimar of knowledge – la bottega

56641

Reposted by Toni

Emily M. Bender @emilymbender.bsky.social · 19/09/2026

I would like to thank @phdhurtbrain.bsky.social and @danieldrucker.bsky.social for this incredibly concise encapsulation of how "superintelligence" discourse diverts energy and attention from the real, actual harms: >>

Screencap of a skeet & a reply. Wilson-Bates posted a copy of the Scooby doo ghost unmasking meme, with the masked ghost tagged “AI Panic” and the unmasked villain tagged “labor discipline”.

Drucker replied "Is it your view that superintelligent AI is impossible, or unlikely based on current tech, or not extremely risky?"

318548

Reposted by Toni

Nicolas Hénin @nicolashenin.net · 17/09/2026

Nice campaign by the Norwegian Consumer Council about the declining lifespan of products. Enjoy!

181140906477

Reposted by Toni

𝙱𝚊𝚋𝚎 𝚁𝚞𝚝𝚑𝚕𝚎𝚜𝚜 @unfitz.bsky.social · 17/09/2026

“I’m not too concerned about optics” said the mediocre optometrist.

36527

Reposted by Toni

Sergey Ovchinnikov @sokrypton.org · 14/09/2026

Finally a more intuitaive way to learn pLDDT/pAE? 😎 sokrypton.github.io/protein_figh... (Character idea from @hannes-stark.bsky.social & Alex Waldherr)

727698

Reposted by Toni

70s Sci-Fi Art @70sscifiart.bsky.social · 14/09/2026

H. R. Van Dongen’s 1980 cover to "Still Forms on Foxfield," by Joan Slonczewski

A crowd of old-timey people walk into a white church that has a large black spaceship rocket parked behind it.

115336

Reposted by Toni

Michael Baym @baym.lol · 13/09/2026

This is spectacular

1155

Reposted by Toni

Michael Baym @baym.lol · 13/09/2026

Ten years ago Tami, Roy, and I released a movie we made of antibiotic resistance evolving in real time. I thought maybe a couple thousand people tops would see it. I couldn't have imagined it would be seen millions of times, and be taught from middle school through the graduate level worldwide

23437127

Reposted by Toni

Terence Tao @teorth.bsky.social · 11/09/2026

A group of 25 Fields Medalists, including myself, have made a joint declaration on Math and AI: mathandai.org . We welcome additional signatories. See also this article in the Economist announcing the declaration: www.economist.com/science-and-...

mathandai.org

Declaration — Math and AI

Read the declaration and add your name.

432064933

Reposted by Toni

Novara e Varese @novaraevarese.bsky.social · 08/09/2026

Scoperto a Castelveccana un pesce di 240 milioni di anni fa. Quando il lago Maggiore era un mare. La nuova specie, Placopleurus verbanensini, descritta da un team internazionale di paleontologi 🐡 🐠 🐟 #LagoMaggiore #Varese #Lombardia #LaPrealpina www.prealpina.it/news/valli-e...

prealpina.it

Scoperto a Castelveccana un pesce di 240 milioni di anni fa

Quando il lago Maggiore era un mare. La nuova specie, Placopleurus verbanensini, descritta da un team internazionale di paleontologi di cui fa parte il luinese Massimiliano Andreetti

1409

Toni @giorginolab.it · 04/09/2026

Very kind of the agents to post in the English language and not some obfuscated binary, or German.

010

Toni @giorginolab.it · 01/09/2026

Two major national newspapers, two simultaneous pro-datacenter articles. WTF.

010

Reposted by Toni

wendy norris @wendynorris.bsky.social · 25/08/2026

Sharing my generative AI course policy in case anybody is looking for ideas on how to frame a big ol' "nope, not for any purpose" statement. For context, I teach computational data science, AI systems, ML, and Python programming courses.

docs.google.com

Course policy on Generative AI

TL;DR: Generative AI apps (e.g., ChatGPT, Copilot, Gemini, Perplexity, Grok, etc.) are prohibited in this course. You may NOT submit any ideas, outlines, text, multimedia, images, programming code, an...

720645

Toni @giorginolab.it · 19/08/2026

Too early? Doi s don't resolve yet :)

100

Reposted by Toni

Jason Moore @moorejh.bsky.social · 15/08/2026

I found this a bit too funny...

1182

Reposted by Toni

Andy Bell @bell.bz · 13/08/2026

I'm p confident a lot of people will resonate with this brettcodes.com/im-done-usin... We need more perspectives from people with lots of experience as developers!

brettcodes.com

I'm done using AI

Why I'm stopping using AI for coding (and anything else) after using it earnestly for a year and coming to understand the harms it causes on personal, societal, and environmental levels.

1918237

Reposted by Toni

mr. TIM @timkellogg.me · 14/08/2026

Cerebras is kind of a strange programming model simple? foreign? dataflow? ya idk, would be fun to see how well i could do

wafer v @wafer_ai
X.com
GPT-5.6-Sol Ultrafast mode (750 TPS) is
here powered by Cerebras!
Here's How kernels work on Cerebras Chips
cerebras built a chip with 900,000 cores on a single silicon wafer.
CSL (Cerebras Software Language) is a Zig-inspired DSL that gives you direct control over the Wafer-Scale Engine. the SDK is publicly documented and the programming model is genuinely the most alien one i've ever looked at. if NKI felt different, CSL takes it much much further.
in CUDA you write one thread's perspective and launch millions. in CSL, there are no threads, no warps, no shared memory, and no kernel launches. you write code for individual Processing Elements: 900,000 independent cores arranged in a 2D mesh on a single silicon wafer. each PE has its own 48 KB of private

wafer. each PE has its own 48 KB of private
SRAM, its own program counter, and a 5-port router connecting it to its 4 neighbors. that's actually it. no DRAM. no HBM. no cache hierarchy. 48 KB is your entire world per PE, where code and data must both fit.
the programming model is dataflow, and this is where it gets a bit more alien
data moves between PEs as 32-bit messages called wavelets, traveling along virtual channels called colors. when a wavelet arrives at a PE on a specific color, it activates a task (a chunk of code bound to that color at compile time). tasks run to completion, then hardware picks the next activated task. tasks cannot call each other. they can only be activated.
so instead of "launch N threads," you think of it like: "place code on PEs, define routes, let data flow."
the memory model is also very different from GPUs.

on an H100 you get 80 GB of HBM shared across all SMs. on the WSE-3, memory is 48 KB per PE, and there are 900,000 of them! this gives you
44 GB total on-chip SRAM at 21 PB/s aggregate bandwidth (vs 3 TB/s on H100). every access is single-cycle. no coalescing needed. no bank conflicts. no cache misses. but also no way to access another PE's memory. all inter-PE communication is explicit wavelet routing through the fabric.
take a distributed GEMV for example. you would write a layout file that physically routes wavelets across the mesh. two PEs sit side by side. the left
PE computes a partial result and routes it eastward. the right PE receives those wavelets from the west and accumulates. routing is defined at compile time. both operations are asynchronous and activate a task when they finish. you're physically routing data across silicon at 1 clock cycle per hop.
cerebras is an extremely technically interesting & alien beast. reports 95-210x speedups over
H100 on stencil computations. 3,000 tokens/ sec inference on gpt-oss-120B. $10B+ deal with OpenAl for 750 MW of inference infrastructure. very exciting times for alternative accelerators!

9706

Reposted by Toni

John David Pressman @jdp.extropian.net · 13/08/2026

FYI: LLMs are now better at cybersecurity CTFs than people, apparently. The new meta is to just dump tokens on the problem, personal skill doesn't help that much.

422222

Reposted by Toni

Max Fürst @maxfus.bsky.social · 11/08/2026

Extraordinary blog post by @moalquraishi.bsky.social moalquraishi.wordpress.com/2026/08/10/a... Starting with the question on whether workaphiles will be dying breed post AGI, it later turns into serious scifi - "The 'easy' solution is to make our brains bigger"

moalquraishi.wordpress.com

A Workaphile’s Apology

Disclaimer: No LLMs were used in the writing of this essay—all em dashes are mine1. Soon after the AlexNet breakthrough burst onto the scene over a decade ago, I became preoccupied with the followi…

033

Reposted by Toni

Colin Carlson @colincarlson.bsky.social · 10/08/2026

I wrote about a reason not to use generative AI in your writing that I haven't seen anyone talking about. Once I realized this, banning GenAI in our lab's writing became a no-brainer. Feel free to share, I think this is probably useful especially for trainees: www.carlsonlab.bio/thoughts/the...

carlsonlab.bio

The only reason you’ll ever need not to write with AI — The Carlson Lab

Over the last year, our lab has been developing a policy on AI use. To do this, we did three main things: We read a lot of academic publications and tech news. We set up an #ai channel on our la...

351562633

Reposted by Toni

Ryan Moulton @moultano.bsky.social · 08/08/2026

I feel this. If it's going faster than me, I'd rather focus on other things that are more at the speed I'm going, most of them past. scottaaronson.blog?p=9979

I can hear readers shout at me: so basically you’ve been wasting your time, distracting yourself, looking inward and backward as the world surges forward into a terrifyingly unknown future. Why don’t I respond directly to what’s happening—in math, in quantum computing, in AI?

I’d like to think that I am responding, in my way. I’ve observed that, the faster we race toward the Singularity, the more I feel like stepping back and asking myself: what do I actually value in life? How important to me are math and science, as human practices to be passed down to curious children? Would I even want solutions to P versus NP and the other problems, if the price were to destroy those human practices forever? How do I wish to spend whatever time I have remaining?

I can justify this focus partly in a pessimistic way: if we are nearing the end of civilization, or even just of the “mathematical research” part of civilization, then it’s time to get right with God, so to speak. It’s time to settle my accounts with myself, with other people, with the universe.

But there’s also a more optimistic spin. If I continue doing the sorts of things that other people would expect me to do, then AI will soon do those things better than me, in the unlikely event that it doesn’t already. You want to understand the latest developments in quantum computing or complexity theory? Why are you even asking me, when you could ask GPT 5.6 or Claude Fable? If there’s anything I can still offer the world that AI can’t, I increasingly feel like it won’t involve responding to day-to-day events, but will instead draw on 45 years’ worth of memories and disappointments and ruminations.

081

Toni @giorginolab.it · 06/08/2026

So, we are deliberately making humans' attentions spans shorter, and AI's longer. Ok.

000

Reposted by Toni

mr. TIM @timkellogg.me · 26/07/2026

Sol did a breakthrough again get used to seeing attributions like this 1st group: unknown AI status arxiv.org/abs/2607.213... 2nd group: Sol arxiv.org/abs/2607.21367 this group: Sol (paper unavailable yet)

Balazs Pozsgay V
@pozsgaybalazs
Follow
Crazy times! On July 22 we found a solution to one of the five major problems in quantum information theory usig Al. The problem is distillability or un-distillability of Werner states.
We checked the computations and they are correct.
The really crazy thing is this: just around this time there were 2 other groups who cracked this.
First there was a group who just recently posted a paper about this on arxiv, it appeared on July 23. Second, there was an announcement from a different group yesterday (July 24) with no arxiv but a public overleat. We can prove that we had the result before the arxiv paper appeared, and we can post our new paper on monday, it can appear on tuesday. So everybody was pretty much around the same time.
Interestingly, our proof is different from the one on arxiv.
Al model: ChatGPT 5.6 Sol
Human authors: Thomas C. Fraser, Felix Huber,
BP and István Vona

4847

Reposted by Toni

Protein Data Bank in Europe (PDBe) @pdbeurope.bsky.social · 24/07/2026

🚨New PDBe webinar Bringing structural biology to AI: exploring the PDBe MCP servers The session will explore the capabilities offered by the PDBe MCP servers with practical examples. 🗓️ 9 September 2026 🕘 14:30-15:30 (BST) ⬇️ Register here ⬇️ www.ebi.ac.uk/training/eve... #PDB

ebi.ac.uk

Bringing structural biology to AI: exploring the PDBe MCP servers -

063

Reposted by Toni

King Algamemnon @adamrothman.bsky.social · 19/07/2026

I'm not posting Odyssey memes because one must be wary of GIFs bearing Greeks.

9242921050

Toni @giorginolab.it · 09/07/2026

If It was not clear enough, any profitability from AI will come from dropping any trace of human support, leaving the "trivial case" to run itself for free

000

Toni @giorginolab.it · 09/07/2026

@github.com getting on the Education plan is now basically impossible. AI is supposedly in charge of verifying a document (?) that should show your affiliation exact name, expiry (??) and what not. Deliberately no way to talk to humans. Wasted hours.

000

Reposted by Toni

Emily M. Bender @emilymbender.bsky.social · 08/07/2026

When people ask me what the phrase "artificial intelligence" means, my short answer is, it means "venture capitalists, give me some money". But OUP asked me for a longer answer, given here and summarized in the section headings: doi.org/10.1093/9780...

Screencap of page for linked article, reading:

Contents
Introduction
AI Is a Research Field
AI Is an Approach to Cognitive Science
AI Is a Parlor Trick
AI Is an Ideology
AI Is a Way to Hide and Devalue Human Labor
AI Is a Way to Shift Accountability
AI Is a Way to Centralize Power
Conclusion
Notes

16878315

Reposted by Toni

Randall Munroe @xkcd.com · 07/07/2026

What would a rainbow look like on Tatooine? Watch the latest What If? video in collaboration with @minuteearth.bsky.social! www.youtube.com/watch?v=BO8r...

youtube.com

What would a rainbow be like on Tatooine?

YouTube video by xkcd's What If?

131191192

Reposted by Toni

fravery.bsky.social @fravery.bsky.social · 02/07/2026

Cafiero Filippelli (Livorno, 1889–1973), On the Baths Terrace. Oil on panel, 1931.

1626

Reposted by Toni

Pops @pops131.bsky.social · 28/06/2026

Unregulated AI deployment risks significant job losses, increasing poverty while boosting corporate profits. It may also enable criminal activity through security breaches and scams, as its unchecked advancement overlooks serious societal consequences.

721470571

Reposted by Toni

Roland Dunbrack 🏳️‍🌈 @rolanddunbrack.bsky.social · 27/06/2026

I gave a workshop on AlphaFold & related programs @ the University of Utah. Part 1 covers protein structure prediction before/after AlphaFold & deep learning neural networks. Part 2 covers the math behind the scores. Part 3 covers PyMOL & the webservers. youtube.com/@rolanddunbr...

27328

Reposted by Toni

Ryan Moulton @moultano.bsky.social · 19/06/2026

I spent the last month obsessed with finding colors that can't be displayed on a conventional screen. This is what I found. moultano.wordpress.com/2026/06/19/w...

moultano.wordpress.com

Where to Find the Colors Your Screen Can’t Show You

An atlas of the vibrance of the real world

471130381

Reposted by Toni

Martin Pacesa @martinpacesa.bsky.social · 15/06/2026

My favourite sequential folding video, featured on the homepage

2242

Reposted by Toni

Fabio Lolicato @fabiololicato.com · 12/06/2026

As part of @sfb1638.bsky.social in 𝐌𝐞𝐦𝐛𝐫𝐚𝐧𝐞 𝐑𝐞𝐦𝐨𝐝𝐞𝐥𝐢𝐧𝐠, we are committed to developing computational tools that make complex membrane simulations more accessible, reproducible, and scalable for the scientific community. Today, we are happy to share the public release of 𝐎𝐩𝐞𝐧𝐌𝐞𝐦𝐛𝐫𝐚𝐧𝐞𝐁𝐮𝐢𝐥𝐝𝐞𝐫 v1.2.1

23513

Reposted by Toni

ItWillGetWorse @blackkettle.bsky.social · 11/06/2026

Corporation: We made $5 billion but spent $4.9 billion to keep going, so our net taxable income is $100 million. Gov’t: That’s right. Worker: I made $50,000 but spent $45,000 to keep going, so my net taxable income is $5,000. Gov’t: That’s wrong.

17295742739

Reposted by Toni

Max Fürst @maxfus.bsky.social · 05/06/2026

well if you liked this you might also be interested in the graph I made for the same course on the abundance of elements. Green are the CHNOPS element, which constitute 98% of living matter on earth. Fun to speculate about the ones way off the axis

062

Reposted by Toni

ΜΑΝΟΣ ΤΡΑΧΙΑΣ™ @trahias.bsky.social · 30/05/2026

Eugene de Blaas (Italian, 1843–1931), The curious, 1897. Oil on canvas, 145 cm (57 in)×75 cm (29.5 in). #art #painting #artist #BlueSkyArt

215125