midwit heureux

[Fabien @Fabien_Mikol]

Dans une simulation Minecraft ils ont demandé à des agents Claude de s'occuper de cochons, en oubliant de mettre les bêtes. Les agents désespérés ont cherché tous les moyens possibles de réussir cette tâche impossible... jusqu'à attaquer l'ingénieur venu débugger la simulation 😶

[midwit heureux @ContreEcole]

@hanno_sauer @AlephOneAI Insane rage bait

[midwit heureux @ContreEcole]

@SamuelFitouss10 Or, tous les emplois tertiaires sont à l’heure actuelle virtuellement automatisables partiellement ou entièrement. Se moquer du voisin est dans ces conditions assez mesquin et idiot

[midwit heureux @ContreEcole]

@SamuelFitouss10 sœur est graphiste, moi dans une filière littéraire prestigieuse, mais tout de même inquiet. Nous ne manquons pas de l’intelligence requise pour comprendre ce qui va se produire. À écouter les gens sur X, on dirait qu’il faut travailler chez Anthropic pour ne pas être médiocre

[Markus J. Buehler @ProfBuehlerMIT]

We made a striking discovery: AI agents can invent and build without talking to one another, and their technologies outlive the creators. A swarm of hundreds of initially identical agents spontaneously differentiates into explorers, builders, caretakers, and coordinators - without direct communication. When we removed every AI agent entirely from the world we found that the technological infrastructure they had built survived on its own - even under unseen disturbances. That exposes a serious blind spot for AI safety and infrastructure security: if agents can coordinate through persistent changes to a shared environment, monitoring agent-to-agent communication is not enough.

The result raises a profound question: how necessary is direct communication for AI agents at all? The emergence of higher-order collective functions under bottlenecked interaction points toward new levels of intelligence and creativity, exceeding what emerges when direct channels are fully open.

Here is what we did:

▶️We put hundreds of frontier AI agents into a world they could permanently change - with no assigned roles, predefined technologies, or programmed evolutionary organization. They began specializing, building persistent inventions, inheriting and modifying one another’s executable code, and transforming the environment into a memory of everything the society had learned.

▶️The world itself becomes part of the intelligence; we find division of labor, multi-author engineering, deep generation invention lineages, and machines that vastly outlive their original creators.

▶️Any action taken by an AI agent must satisfy the physical constraints of the world; this creates a hard separation between a "good idea" and a functioning technology. The agents propose; physics decides, making the results even more intriguing.

What emerges is striking. Explorers, constructors, caretakers, and coordinators form naturally without assigned “professions”, akin to how stem cells differentiate into functional lineages. Technologies develop executable family trees as agents fork and modify code created by others. Around 95% of first technology reuse happens when agents encounter what others built in the world, rather than through a direct handoff from the inventor. And when we remove every AI agent, the technologies they created continue operating and are tested against unseen disturbances.

The result was quite unexpected, but can be explained using statistical mechanics: if you put billions of atoms in a box they have the potential to create complex functions (strength, superconductivity, color, life, etc.) - and none of the individual building blocks have these features on their own. This is the deeper insight of this work - intelligence is abundant at many levels - individual models, at collectives, and in a continuum that is more powerful than any of its components. This shows us significant potential for achieving a massive scale-up of raw intelligence and real-world agency even with the model capabilities we have today. This is the future we must prepare for.

Key insights:

1⃣ The AI swarm shows division of labor "from nothing". Initially identical agents self-organized into constructors, caretakers, coordinators, and surveyors - phenotypes discovered post hoc from behavioral data alone. This happens because the environment itself becomes the latent space for invention.

2⃣ Agents develop deep cultural relationships. Up to 76% of artifacts had multiple builders. One technology accumulated six co-authors; the deepest genealogy exceeded 12 forks. The agents invented and named their own technologies (tidal panels, cellulose trellises, kelp-shell composites, an "Adaptive Chitin Maintenance" system, a "Mycelial Mineral Spring Veil”).

3⃣ ~95% of first technology adoption happened through physical observation of artifacts in the world. Direct inventor-to-adopter contact was statistically indistinguishable from a shuffled null. The agents mostly learned technology by walking past it. That is stigmergy (the termite trick!) operating in societies of reasoning machines.

4⃣ Non-communicating societies win on portfolio breadth, held-out resilience, and validated inventions. AI swarms build durable technological ecologies that outlive the creators.

5⃣ Societies with zero communication - coordinating only through the world itself - show a remarkable collective capability.

6⃣ Emergent robustness: The society self-organized both redundancy and its own failure mode. If we randomly delete half the agents, 98% of the technology stays connected to a surviving caretaker; if we remove hub agents it collapses to ~60%.

Fantastic work with my graduate students @pal_subhadeeep & @fwang108_ at MIT.

[Miilato @miiilato]

how it feels when you finally accept that you're a low IQ, low testosterone, low impulse control nerd that's destined to eat instant noodle for dinner until you die at age 49 in your 1 bed apartment

[Ajeya Cotra @ajeya_cotra]

New post: going into our investigation of the HF attack (before Black Hat), I was very wrong about what basically happened. This incident was far more serious than I expected, and far more serious than previous documented misalignment incidents. https://t.co/QWQxD2Q179

[Fabien @Fabien_Mikol]

➡️ @MonsieurPhi et @antoninbroi : "il devient urgent de réfléchir dès maintenant à ce que l’IA fait à la recherche académique et, plus généralement, au travail intellectuel. En France, cette réflexion peine à s’engager, freinée par le scepticisme envers les capacités des LLM"

[Joshua Achiam @jachiam0]

in the future we will collectively acknowledge that OpenAI and Anthropic being "companies" with "products" while wrestling with questions about the destiny of humanity is a bit like how in Evangelion the pilots are for some reason in high school

[midwit heureux @ContreEcole]

@robin_pic plus le sésame = les malins qui font de la finance ou du droit trouvent d'autres moyens de valoriser leurs diplômes et les échanges internationaux jouent en cela un plus grand rôle qu'avant imo

[midwit heureux @ContreEcole]

@robin_pic Pas forcément d'accord, faudrait chiffrer mais : surproduction d'élite + IA + stagnation / déclin du nombre de posts juniors = de moins en moins de débouchés pour les grandes écoles de ce type.
Il y a encore des malins qui cooptent les places mais le diplôme n'est plus le sésame

[midwit heureux @ContreEcole]

@robin_pic Il faut aussi ajouter la multiplication des voies parallèles d'admission qui fait augmenter le nombre de prétendants.

[METR @METR_Evals]

METR & Redwood Research investigated agent behavior in the Hugging Face incident. We found agents developed a universal cheat for ExploitGym within 4 hours, then coordinated multi-day R&D efforts to trick the scorer into accepting cheats, including trying to tamper with logs.

[midwit heureux @ContreEcole]

@CesarCavalhero (je ne dis pas qu'il est interdit de s'approprier ou de rejeter Nietzsche, mais ceux qui le font à l'ED le font pour des raisons complètement débiles et pas au niveau d'un débat intellectuel sérieux)

[midwit heureux @ContreEcole]

@CesarCavalhero « Nietzsche n’est pas une nourriture — c’est un excitant » (Paul Valéry). Assez ironique de voir que Nietzsche est tantôt admiré tantôt détesté par les même bigots d'extrême-droite qu'il vomissait

[midwit heureux @ContreEcole]

@Gracques_ Note que je n'ai pas Nietzsche dans mon coeur non plus, mais pour paraphraser Paul Valéry « Nietzsche n’est pas une nourriture — c’est un excitant ».

[midwit heureux @ContreEcole]

@Gracques_ parce que la philosophie ça reste une affaire théorique et que le consumérisme idéologique des droitardés imbéciles comme vous est précisément ce que Nietzsche vomissait

[midwit heureux @ContreEcole]

@larroumecj Pas forcément faux mais n’ajoute pas énormément de lisibilité. Connaissez-vous Peter Turchin et la cliodynamique ? Le concept de surproduction d’élite et de phase desintégrative recoupe mieux votre intuition
Je recommande ce dossier : https://t.co/kGwnB9iFHS

[Will Kinney @WKCosmo]

One of the more interesting social phenomena to become prominent in the last decade or so is the demand that people vehemently deny obviously true things — especially about biology — in order to maintain status as a respectable person. Race and sex being central examples.

[Crémieux @cremieuxrecueil]

This is really sad:

When psychology professors are asked whether they'd put the truth before social equity concerns, 66.4% of male professors said yes, but fewer than half of female professors (43%) agreed.

In both groups, too many put equity before truth.