George Pór #StandWithUkraine

This is an AI's response to my prompt "technoshaman is looking into the mystery through the eyes of the beloved community." I kind of like it. If we explored the possibility of some kind of collaborative AI artform, what prompt would YOU be interested to have the AI render?

Thought experiment: If you said “I support Israel’s onslaught against Gaza” there would be no consequences for your career. What would happen if you said “I want to do to Israel what they’ve done to Gaza”? Easy way of testing whose lives matter and whose don’t!

@AaronPettman It’s too late to bring the word back into the industrial era, when people didn’t have the tools for taking self-organisation to scale!

"F*ck AI!" AI data centre protest now on Granville Bridge, Vancouver BC

I just bought Pap Szilárd a Coffee on Ko-fi! ❤️ Ko-fi.com/home/coffeesho…

@T_C_S_ @michaelgarfield @neno_is_ooo ditto

@michaelgarfield @neno_is_ooo Inspiring. I’m going to follow this core team and I hope to learn and evolve with you.

I've been quieter than usual the last few months. Here's why: The hundreds of conversations on public record at Humans On The Loop were just the tip of an iceberg. Over the two years since leaving Mozilla, I've met more and more amazing people who share a dream of a future I actually want my kids to inherit: a flourishing world where narrative authorship is restored to the people, ecologically-grounded economic systems are built with care, and sovereign sensemaking infrastructure enables us to meet the wicked problems of our century with the creativity and collective intelligence they require. Atlas Research Group, the team I've been working to help launch, is a kind of seed crystal for a federated commons of open source projects in service of this future. My colleagues on the core team — @dylantull, @coreygo, @csageland, @QBFrank, and @HexaField — are some of the most insightful and capable people I've ever had the pleasure to team up with. All of us were working toward this for years without realizing just how many of us there are. And now we're starting to fold in even more incredibly brilliant, caring, and talented people — all of us composting our projects into this bigger thing that finally feels like it's speaking itself into being. I'll have much more to say about this later, but for now I'm just happy to announce our introductory article where we explain what we're about and extend your invitation. You'll find a link to the whole thing and an especially juicy excerpt below:

You will know that the AI labs believe in ASI when they disband their newly formed consulting (sorry “forward deployed engineering”) groups. As long as people are required to figure out how AI is useful & do organizational change & systems integration, jobs seem to be pretty safe

@srlevons @marcarza Does anybody remember of the song from the 1970s by Pete Seeger , Little boxes on the hillside, little boxes… ?

@marcarza Hauria de perdre’l abans aquest atractiu. Als 40 la gent ja hauria d’estar amb una estabilitat i en família majoritàriament. Però per molts motius, principalment culturals però també econòmics estem allargant l’adolescència.

L'habitatge en propietat és una palanca de llibertat, d'arrelament i de seguretat. La vida mòbil i inestable del lloguer etern perd molt d'atractiu a partir dels quaranta anys.

Happy International Workers’ Day! To kick it off, here’s a thread of #MayDay posters from around the world. Let's start in Austria, with this poster from 1891.

Holy shit... Andrej Karpathy tweeted he wanted LLM-powered knowledge graphs. 48 hours later, someone shipped it on GitHub. It's called Graphify. One command. Any folder. Full knowledge graph. 71.5x fewer tokens per query vs reading raw files. No vector DB. No embeddings. No config. Point it at a folder of code, papers, PDFs, screenshots, even whiteboard photos in other languages. It reads everything, builds a queryable graph, and hands you back structure you didn't know was there. You can ask it stuff like: → "What calls this function?" → "What connects these two concepts?" → "What are the most important nodes in this project?" Out the other side you get: → A navigable interactive graph (click nodes, search, filter) → An Obsidian vault with backlinked articles → A Wikipedia-style wiki that starts at index.md → Plain English Q&A over your entire codebase Two-pass pipeline. First pass is a deterministic AST extraction across 19 languages via tree-sitter (Python, Rust, Go, TS, Swift, Zig, Elixir, the works). Second pass runs Claude subagents in parallel over docs, papers, and images to pull out concepts and design rationale. Leiden community detection clusters everything. SHA256 cache means re-runs only touch changed files. It even installs a git hook. Every commit auto-rebuilds the graph. Every branch switch too. The wildest part? The benchmark folder is literally Karpathy's own repos. nanoGPT, minGPT, micrograd, plus the Attention Is All You Need paper and FlashAttention 1+2. Graphify found that nanoGPT Block and minGPT Block are linked across repos, with the FlashAttention paper bridging into CausalSelfAttention in both. That's the structure no keyword search will ever give you. 6.3k stars in days. 100% Open Source. (Link in the replies)

The Russians know that if we withdraw from Donbas, it will divide our nation. Many people would be against it. If Ukraine’s unity is broken, Putin will come up with a new “blitzkrieg” to occupy Ukraine very quickly. Even if he doesn’t, he will use a subsequent pause to recruit more people, train them, ramp up the defense-industrial base, and get sanctions lifted. So a ceasefire in the format “stay where we stay” is not only our wish, it is in the interest of our partners too. From an interview on The Rest Is Politics podcast (3/5).

@AryJeay The same people who called you #antisemite for opposing #GazaGenocide‌ have bombed Jewish synagogue in Iran 🇮🇷

Anyone Who Tells You That We Can't Put Criminals in Jail Until We Win an Election Is Complicit in the Murder of Our Democracy open.substack.com/pub/cmarmitage…

This is buried under the noise. By November, it may be the story that mattered most. If you read one thing today, make it this.

China just dropped its 15th five year and I genuinely think this is the most important document of 2026 & almost nobody in the West is talking about it what gets me isn’t any single number…it’s the systemic vision, in one document they’re planning GDP growth, highvalue patents, digital economy at 12.5% of GDP, non-fossil energy at 25%, CO2 reduction, surface water quality at 85%, life expectancy at 80, forest coverage rate, grain production capacity…ALL of it in the same table with hard targets & binding indicators through 2030 in the West we treat economy, ecology, education, healthcare & energy security as separate ministries fighting over budgets, China treats them as organs of the same body, every line feeds the others, urbanization fuels productivity, R&D fuels patents, patents fuel the the digital economy, non-fossil energy fuels strategic independence….every variable is designed to accelerate all the others & that reflect china’s deeply HOLISTIC worldview look at line 5, high-value patent going from 16 to over 22 per 10,000 people by 2030, quietest line in the whole table and probably the most important, a high value patent is the intellectual property of a technology that will structure a market for 20y MINIMUM China is literally planning who owns the technological future, the licenses & dependencies other countriees will rack up on Chinese tech, this IS the economic and industrial colonization of the 21st century I keep talking about in my threads but with patents instead of cannons while america bombs & europe debates, China plans, builds & executes and the scariest part is that based on their own audits the previous plans hit ove 90% of their targets Western leaders rotate every 4 /5 years and make promises they’ll never have to keep, CCP plays on a 50 y horizon…this table isn’t a campaign platform, it’s a national spec sheet & that changes everything about how you read each number in it

@IliasAlami You can watch the interview here, you can switch between HU and EN audio tracks in the Settings. youtu.be/6q3GlfEJGDE?si…

@AnthropicAI Will the Institute ever transcend its defensive posture of talking only about the dangers of advanced AI and start exploring the potential of AI to help us discover the paths to a regenerative future?

Introducing The Anthropic Institute, a new effort to advance the public conversation about powerful AI. anthropic.com/news/the-anthr…

@emollick But if your main interest is business benefits, then think of it as the ultimate, uncatchable strategic advantage!

@emollick How can AI help us deepen our individual and collective wisdom is much more interesting and useful than taking AI only for a productivity or efficiency coach.

The "how does AI improve individual productivity" discussion is much less interesting than "how does AI improve organizational efficiency?" We have a lot of answers to the former, but it will always get eaten by the latter if we don't experiment with new approaches to organizing.

@marcarza La imaginació sense acció és somiar amb els ulls oberts. L'acció sense imaginació porta a malsons. Imagination without action is daydreaming. Action without imagination leads to nightmares. to Marc and @JCartanyaReus

Els nostres problemes no són de gestió, són d'imaginació i d'acció. No és temps per tecnocrates, ni aquí ni enlloc, és temps d'imaginòcrates.

@PulaRJS @markcurtis30 40 (imaginary) dead babies filled every headline in the West on 8 October 2023, but 127 (real - alive yesterday) dead schoolgirls won’t rate even one Western headline today.