Marcus van der Erve✨

@NguyenVero54460 Sarah Knafo -> European Javier Milei

Geoffroy Lejeune : « Sarah Knafo, pourquoi ça marche dans l'opinion ? Parce que les gens se rendent compte que la baisse des dépenses de l'État, comme elle le propose, leur évitera des impôts supplémentaires. »

Paraphrasing Argentina's Milei: "Free markets are the best way to create scientific wealth AND reduce scientific poverty." -> the markets of physics and science are no longer free markets... ...think about that.

@knafo_sarah Sarah Knafo (Milei)... 👍🏻

Dépenser moins pour taxer moins, ce n’est pas qu’un slogan que je répète depuis des mois. C’est concret. Avec une seule mesure d’économie, la suppression de l’aide publique au développement, le prix du carburant baisserait de 24 à 27 centimes par litre. Regardez l’impact sur votre pouvoir d’achat de cette première mesure et imaginez l’argent que vous pourriez récupérer si l’on supprimait toutes les gabegies. ⛽️ Calculez vos économies sur combienjegagne.fr

@latestincosmos Not even that...! zenodo.org/records/178407…

So gravity isn't a force, it's a distortion of space and time. For something to distort it must have a structure, so space and time must have a structure. A structure must be made out of something; be it matter, energy or a fundamental force. So my question is what is this 'space time structure' made out of?

@leap_71 @Fraunhofer Wow 🤯

The next frontier of 3D-printing: multiple different metals. Here is an experimental aerospike rocket engine design we manufactured with the @fraunhofer IGCV bimetal printing process in 2023. One of the first projects we built on our open-source PicoGK geometry kernel.

@MaMoMVPY ASI (Artificial Super Intelligence) hinges on a true sense of spaceTIME and related memory dynamics... no LLM provider has it or knows how to really achieve it other than in a mechanistic, Cartesian way.

I am increasingly coming to this conclusion - there NO SIGNS the problems with hallucinations are getting solved. In fact if anything it is now spreading to coding and the use of agents where it is hidden and could led to serious problems down the road. Therefore we can't really scale LLMs. LLMs are very useful tools, but you need to know about the limitations. Most of the investments in LLMs today assumes away these limitations.

@Kasparov63 @GaryMarcus ASI (Artificial Super Intelligence) hinges on a true sense of spaceTIME and related memory dynamics... no LLM provider has it or knows how to really achieve it other than in a mechanistic, Cartesian way.

These results keep coming. @GaryMarcus was way ahead on calling out these problems and dangers. And we're at risk of baking these issues into systems we rely on.

@haider1 ASI (Artificial Super Intelligence) hinges on a true sense of spaceTIME and related memory dynamics... no LLM provider has it or knows how to really achieve it other than in a mechanistic, Cartesian way.

openai next model being a "new pre-trained base" is the big thing if spud is trained from scratch on 2 years of new data and has a meaningfully different architecture, that would be a real step change. the native omnimodal part matters too. current models added vision later, but native multimodal training is different

@Dr_Singularity ...including @ylecun of course.

@Dr_Singularity ASI (Artificial Super Intelligence) hinges on a true sense of spacetime and related memory dynamics... no LLM provider has it or knows how to really achieve it other than in a mechanistic, Cartesian way.

OpenAI is saying that we are entering the Intelligence Age, and it will reshape the economy much faster than most systems are prepared for. "Now, we’re beginning a transition toward superintelligence: AI systems capable of outperforming the smartest humans even when they are assisted by AI." Right now, the entire system is built around human labor. People work, earn wages, and pay taxes. However, as AI starts doing more of the work, this model begins to break down while productivity continues to grow rapidly. This creates a strange situation where the economy expands quickly, but traditional jobs do not grow at the same pace. OpenAI’s point is not to slow this process down. Instead, they argue that we need to redesign the system around it. This could mean shifting taxation away from wages toward capital and AI driven output. It could also involve giving people shared ownership in the AI economy, building automatic safety systems that respond to disruption, and reducing working hours while maintaining living standards. They also emphasize that access to AI should be widespread, similar to infrastructure like electricity or the internet, because access to intelligence will become a key source of opportunity. This transformation is coming quickly, and the potential benefits will be enormous.

ASI (Artificial Super Intelligence) hinges on a true sense of spaceTIME and related memory dynamics... no LLM provider has it or knows how to really achieve it other than in a mechanistic, Cartesian way.

No sense of spacetime for LLMs yet -> a fundamental model upgrade is needed that has not yet been specced out by any of the LLM model providers.

@haider1 No sense of spacetime for LLMs yet -> a fundamental model upgrade is needed that has not yet been specced by any of the LLM model providers.

Sam Altman says today's models can already help scientists make small discoveries But the next class of models will help produce the most important discovery of a researcher's decade, or even their career "soon, one person with 100x GPUs will do the work of an entire software team"

@bhavaytyagi @seanmcarroll Ontology is, in fact, what bogs physics down to established premises... ...remember that when the pre-metric era slides in...

As someone who deeply thinks about ontology and epistemology in physics, I’m glad this reflection was published and such discussions are happening. In the words of @seanmcarroll (up to paraphrasing), we have no right to understand quantum gravity if we don’t fully understand quantum mechanics. In a recent discussion between Sean and Harlow, it was genuinely nice to see them questioning the foundations of QM to gain insights about gravity from different vantage points.

Most tools help researchers search faster inside an inherited frame. RGPxScientist is built for an earlier step: to help reframe the question itself. You bring a phenomenon, anomaly, or hunch. The app works to clarify what is changing, what is coordinating, what may stay invariant, and what minimal test could show whether the framing deserves to survive. That matters because many fields are not short on answers. They are short on better first questions. RGPxScientist is a front entrance into coherence-first research. chatgpt.com/g/g-695e7b3d73… #RGPxScientist #researchtools #science #AI #complexsystems

CMB may still be read too much as a temperature map and too little as a morphological record. If so, part of the missed signal is not “more energy content” or “one more parameter,” but organization: directionality, persistence, asymmetry, and coherence under load. That is where a pre-metric extension becomes interesting. Not because standard cosmology has failed at everything. But because morphology may carry information that metric summaries compress away too early. A useful next step is simple: treat large-scale patterning not as residue, but as evidence. zenodo.org/records/187599… #CMB #cosmology #physics #RGPx #premetricphysics

Much of science still explains stable forms as if the form were primary. Our work starts one layer earlier. Not with objects first, but with gradients, their recursive coordination, and the conditions under which coherence holds or breaks. That shift matters because many hard problems do not fail for lack of data. They fail because the wrong level of description is treated as fundamental. RGPx asks a stricter question: what must remain coherent for a structure, process, or law-like regularity to persist at all? That is the level where a better bridge may appear between physics, complex systems, cognition, and societal order. #RGPx #complexsystems #physics #coherence #science

@BoWang87 Bio... is on its last substrate leg...

Solving biology should be the ultimate test for AGI 🔥 Anthropic is cooking! Now we need more frontier AI labs to step in—this is where it really matters. The next wave of AI & Bio is going to be wild.