Rafael Kaufmann (🧠,🌎)

@_digitalgaia Last week we shared @_digitalgaia's vision: defeating the Metacrisis by building a global network of regeneration copilots. Now you can help us build it faster today! 1/ Please donate on Gitcoin, there is a $350k matching pool. 2/ Retweet to amplify! explorer.gitcoin.co/#/round/1/0x42…

It's been released just a few days ago and already more than 500 derivative models of @deepseek_ai have been created all over the world on @huggingface with 2.5 million downloads (5x the original weights). The power of decentralized open-source AI!

@lmeyerov @ApacheArrow @marimo_io gaia-lab.de. For code - not yet in public. For theory/vision - this talk by @_rkauf : youtu.be/47r5P6xtjms?fe…

@xuanalogue @_rkauf and I proposed intersubjective Bayesian model network as a basis for such contracts: engineeringideas.substack.com/p/gaia-network…

Last up for this session @_rkauf (with absolutely the best shirt in the room) and Thomas Kopinsky are presenting about "Scalable Safety in Complex AI-Human Systems" - and this type of modelling of complex risks seems useful in identifying safety issues to mitigate.

@davidmanheim @_rkauf

Elon Musk has complied after all and provided a list of X Holding Corp shareholders who helped the billionaire buy Twitter. And who do you think turned out to be the investors of the X platform? That's right, Petr Aven and Vadim Moshkovich. Now it is absolutely obvious where Musk got such love for Russia and where the roots of “filters” due to anti-Russian publications grow from. Let me remind you who Aven and Moshkovich are. Petr Aven is a Russian billionaire, founder of Alfa Group, which is one of the main wallets for Putin. This man is controlled by the Kremlin as much as possible, for example, in 2018 he traveled to Washington with his colleague Fridman to lobby on behalf of the Russian government to lift international sanctions on Russia. Aven is one of Putin's oldest friends. In general, without Peter's interference, Putin would be in jail rather than becoming president. After all, Aven covered for Vladimir Putin in 1992, when he organized the illegal trade in export licenses. Aven promptly approved him as an authorized representative of the Ministry of Foreign Economic Affairs with the right of foreign economic activity, thus saving the future dictator from prison. Aven also attended a meeting with Putin on February 24, 2022 to mark the start of the invasion of Ukraine. He did not express any protest, started the special military operation and silently supported Putin. By the way, he along with Friedman is trying to get sanctions lifted, just yesterday he received another rejection in this matter. Twitter is represented in the purchase by 8VC Opportunities Fund II, L.P., a company in which his son Denis works. Vadim Moshkovich is a sub-sanctioned Russian agrarian billionaire who was a member of the “Federation Council of the Federal Assembly of the Russian Federation”. In other words, he was officially part of Putin's closest circle. Moshkovich owns Russia's largest agricultural holding Rusagro. Naturally, he also attended the meeting with Putin on February 24, 2022 and, like Aven, did not express any protest against Russia's invasion of Ukraine. Twitter is represented in the purchase by 8VC Opportunities Fund II, L.P., a company in which his son Jack (Eugene) works. Also an interesting personality is Al-Waleed bin Talal Al Saud, who both personally and through his Kingdom Holding Company is represented in the list of investors as many as three times. This man continues his investment business in Russia and does not publicly support Hamas in the war against Israel. He regularly works with Rosneft, Lukoil and Gazprom on investments in Russia. Knowing all of this, how can there even be a discussion about Elon Musk's bias? Everyone was thinking how is it possible that Twitter (X)'s stock is plummeting, advertisers are breaking contracts, and no one is doing anything about it. I mean, that's a direct loss to investors. But what everyone overlooked was the fact that investors didn't care about profits, they needed a platform to promote their narratives, such as: supporting Putin and Russia; lobbying to freeze the war in Ukraine; taking the Ukrainian agenda out of the top news; the righteousness of Hamas, etc. It is now obvious that all the changes to Twitter that are ruining this beautiful platform are a consequence of Elon Musk's collaboration with people like Moshkovich and Aven. #ElonMusk #RussiaUkraineWar #RussialsATerroristState

Envisioning a regenerative economy with #AI as the co-pilot, where every stakeholder, from farmers to consumers, harnesses data-driven insights for impactful decision-making. @_digitalgaia @_rkauf 🎧 Full episode: pod.fo/e/179819 /5

@leventov @acsresearchorg @AmmannNora @robertwiblin @AndrewCritchPhD @InferenceActive Fantastic work @leventov

FYI @_rkauf @acsresearchorg @AmmannNora @robertwiblin @AndrewCritchPhD @InferenceActive

Connecting @mushtaqkhan100's institutional economics and political settlement framework with @drmichaellevin's and Chris Fields' regulative development and scale-free morphogenesis and cognitive science theories. open.substack.com/pub/engineerin…

It's just remarkable (and worrying) how business leaders and journalists misunderstand the core issues with AI adoption and transition. All they talk about is "accuracy", "correctness", and "proving that AI is actually right"(!). This "rule-following and accuracy" framing is about greasing the bureaucracy and can lead to Jevons effect (if bureaucracy is cheap now, let's do more of it!) without actually making cooperation and control more reliable. Making AIs use language effectively to communicate with humans and each other is not an "accuracy" or "factual correctness" problem, it's the context (meaning, intent) alignment problem. Preventing outright hallucinations is just the lowest-hanging piece of it. And “aligning LLMs with human values” is hardly a part of it at all. It's richer language, better context alignment, and better capacities for (compositional, collective) reasoning, bargaining, planning, and decision-making that make the economy more productive and civilisation safer at the end of the day, not “better bureaucracies”. In short, AI safety = Context alignment throughout pre-training (ex.: @OpenAI's Superalignment), fine-tuning (ex.: Armstrong and Gorman's Aligned AI), and online inference (who?) + Languages and protocols for effective, precise, and compositional context alignment and (collective) reasoning (coordination, decision-making) about preferences, plans, and risk bounding (ex.: @davidad's OAA, @_rkauf and my Gaia Network agenda, Friston et al.'s "Ecosystems of intelligence", @ConjectureAI) + Provably correct computing (@tegmark and @steveom's "Provable safe systems") + Governance mechanisms, incentive design, and immune systems to negotiate and encode high-level, collective preferences, goals, and plans and ensure that the collective sticks to the current versions of these (ex.: @collect_intel, @AIObjectives, @drmichaellevin and Chris Fields' regulative development agenda, etc.)

@leventov @davidad @AndrewCritchPhD @hendrycks @jankulveit @_ramsaybrown @AIObjectives

@_rkauf @davidad @AndrewCritchPhD @hendrycks @jankulveit

Gaia Network: a practical, incremental pathway to Open Agency Architecture, written with @_rkauf lesswrong.com/posts/AKBkDNeF…

@CityofHoboken @thehobokengirl @RonBautistaNJ @RaviBhalla When will this practice stop? We are in 2023, we have VisionZero, electric EV stations, CitiBikes, and progress. But EVERY FREAKING FRIDAY the streets are littered ALL around. Bad for sewers, flood, pollution, and trees.

Hey @X (or @Support) 👋 It seems you made a mistake by removing @stephnass. He's all but a spammer. He's the founder of @OpenVC_. Please fix this mistake and reinstate his account 🙏

@gaiaconsortium's Gitcoin grant is finally out! Help us create a regenerative AI network to overcome the Metacrisis! #GR18 explorer.gitcoin.co/#/round/10/0xb…

Final session, the Roundtable discussion, about to begin! youtube.com/watch?v=PVeyvH… With: Karl Friston, Anna Lembke, Rafael Kaufmann @_rkauf , Bert de Vries @bertdv0 , Guillaume Dumas @introspection , Curt Jaimungal @TOEwithCurt Program: @active-inference-institute/3rd-applied-active-inference-symposium/program-4" target="_blank" rel="nofollow noopener">coda.io/@active-infere…

oh no, please don't pit my climate gang against my watershed restoration gang, it's tearing my family apart

We are excited for the upcoming 3rd Applied Active Inference Symposium at the Active Inference Institute: “Enacting Ecosystems of Shared Intelligence”, on August 22nd. The program is now completely formed and looks incredible: @active-inference-institute/3rd-applied-active-inference-symposium/program-4" target="_blank" rel="nofollow noopener">coda.io/@active-infere… We’ll be hearing many voices from the Active Inference Ecosystem, including: Karl Friston, Inês Hipólito @ineshipolito, Aswin Paul @aswinpaule, Takuya Isomura @TakuyaIsomura, Shanna Dobson, Nynke Boiten, JF Cloutier @jfcloutier, Conor Heins @conorheins, Bert de Vries @bertdv0, Dmitry Bagaev, Bart van Erp, Sanjeev Namjoshi, Rafael Kaufmann @_rkauf, Avel Guénin-Carlut @FluffyCyborgII, Pablo Fernandez-Maquieira @Pablo_FMM, Mahault Albarracin @MahaultAlbarra1, Anna Lembke, Guillaume Dumas @introspection, and Curt Jaimungal @TOEwithCurt. Register at: @active-inference-institute/3rd-applied-active-inference-symposium/registration-2" target="_blank" rel="nofollow noopener">coda.io/@active-infere… There will be two intervals of the Symposium on August 22nd, 2023. Both intervals will be livestreamed to YouTube, and watchable/re-watchable at the same link. The first interval will begin at 2 UTC youtube.com/watch?v=rIemcs…, and the second interval will begin at 14:30 UTC youtube.com/watch?v=PVeyvH…. Following the second interval, we will continue the discussion on the Institute’s Discord. Feel free to join the Discord now to start connecting with the community and engaging with Institute activities @active-inference-institute/aii-onboarding/activities-7" target="_blank" rel="nofollow noopener">coda.io/@active-infere… Let us know here or by email if you have any questions.

@desamtralized 🙏🙏🙏💚💚💚

GM Today is the birthday of one of the brightest mind I had the pleasure to work with on this earth: @_rkauf Wish him happy birthday and many years!

@conorheins @BerenMillidge @lancelotdacosta @richardpmann @icouzin @pgupta_ci @jacoblijie BTW, time flies! 3 years later, the methods in that paper would strongly benefit from an upgrade to modern JAX-based probabilistic programming, as we've been using in @_digitalgaia and it looks like you guys have been doing the same :)

@conorheins @BerenMillidge @lancelotdacosta @richardpmann @icouzin Excellent work @conorheins @BerenMillidge and team! Follow-on work could study the development of cog capabilities (ex: Theory of Mind, goal alignment) as an explicit bridge between individual and collective ActInf. Cf @pgupta_ci @jacoblijie & me, 2021 mdpi.com/1099-4300/23/7…

New preprint! “Collective behavior from surprise minimization,” by me, @BerenMillidge, @lancelotdacosta, @richardpmann, Karl Friston, @icouzin Paper: arxiv.org/abs/2307.14804 Code: github.com/conorheins/col… 🧵 (1/14)