[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

@RomainPete98794 Avatar @RomainPete98794 Romaiⁿ ⊶ Peter

Romaiⁿ ⊶ Peter posts on X about ai, not in, elements, lever the most. They currently have XXX followers and XX posts still getting attention that total XXX engagements in the last XX hours.

Engagements: XXX #

Engagements Line Chart

Mentions: XX #

Mentions Line Chart

Followers: XXX #

Followers Line Chart

CreatorRank: XXXXXXXXX #

CreatorRank Line Chart

Social Influence #

Social category influence countries finance fashion brands

Social topic influence ai, not in, elements, lever, france, germany, proof of, lack of, clarity, gpt5

Top Social Posts #

Top posts by engagements in the last XX hours

"Do you mean that transitivity-based thinking (that's what A C B is) should be considered thinking rather than pre-existing knowledge if it's not in the original corpus of knowledge but obtained by the LLM itself by the combination of elements of his knowledge corpus That's literaly one of the most elementary forms of thinking (aside contradiction and implication) so that's definetly thinking. Is this all that reasoning can be No there are stronger forms of reasoning that can't be captured by this kind of logical inferences (I explained why in another post link below)"
X Link @RomainPete98794 2025-11-03T10:29Z XXX followers, XX engagements

"@Vinci_TreeEye"
X Link @RomainPete98794 2025-11-03T10:29Z XXX followers, XX engagements

"I can't. All I see is abstractions now. Mushrooms are abstractions too. It is not the concrete fungal chemistry with which you discredit me through your mockery; it is the abstract idea of a hallucinogenic power corrosive to my faculties of judgment. The mushroom as an abstract power of irrationality. 🍄"
X Link @RomainPete98794 2025-10-31T19:04Z XXX followers, XX engagements

"France took the lead even before the war. In the early 1930 brillant French mathematicians went to Germany mostly using Rockfeller grants to study German structural math. In the mid-1935 they came together and founded Bourbaki that would carry the flame of structural mathematics for decades. Some of their important works were published before or even during WWII"
X Link @RomainPete98794 2025-11-01T04:13Z XXX followers, XX engagements

"@NoahChrein The same can be said of "fast-track human race annihilation". Seems to be a proper eschatological telos"
X Link @RomainPete98794 2025-11-03T11:15Z XXX followers, XX engagements

"@repligate If the prompter really care about information density and compression its model will also. The original post isn't a universal statement it's a performative proof of a lack of AI ability"
X Link @RomainPete98794 2025-11-04T05:24Z XXX followers, XXX engagements

"Very impressive Some thoughts: a significant aporia exists between the theoretical equivalence of the Lagrangian Eulerian and Semigroup conditions (Proposition 1) that you state and their stark empirical divergence in practice (for example G-LSD's consistent superiority ). The paper relegates this critical gap to "future work" and that may be one of the most important ones. Where precisely does this equivalence break down Is it in the discretization The specific stop-gradient implementation Or does the divergence reveal a fundamental flaw in the very premise of their equivalence when moving"
X Link @RomainPete98794 2025-10-30T16:44Z XXX followers, XXX engagements

""If you cant define then it doesnt exist" .That's quite a stupid axiom. Conflates existence with knowledge. Conflates the building process with with the clarity of the telos. Definition can sometimes be attained only when you've built something. Sometimes the depth of definitions is greater that your explicit knowledge (that what you discover by doing analytic reasoning of proper defined objects in math). Things are WAY more complicated than this axiom"
X Link @RomainPete98794 2025-11-03T11:33Z XXX followers, XX engagements

"Maybe the wrong way is to call it "prompt" engineering. You're not engineering the prompt you're engineering the correlations on the latent space. The prompt is not the object of engineering and as long as you think this you miss the real object of the intervention. When you think that way you can even start to go beyond existing correlations to engineer a new geometry for the semantic space by engineering the metrics of the space that are relevant to your goal. The "prompt" is a mere tool"
X Link @RomainPete98794 2025-11-03T21:15Z XXX followers, XX engagements

"If anyone has GPT-5 Pro requests to spare I can send a full prompt to test"
X Link @RomainPete98794 2025-11-04T10:44Z XXX followers, X engagements

"Is there one more step forward Moving from "prompt" and "context" engineering; not seeing the object of engineering as things human write that can be read. Thinking you're not engineering the prompt or the context you're engineering the correlations on the latent space. When you think that way you can even start to go beyond existing correlations to engineer a new geometry for the semantic space by engineering the metrics of the space that are relevant to your goal. You're not curating and optimizing content anymore you're forcing the model on live in ontologies where semantic distances are"
X Link @RomainPete98794 2025-11-04T13:07Z XXX followers, XXX engagements

"I've continually hit the ceiling of mathematical imagination for generating theorematic gestures for serveral weeks now. So I decided to be more radical in my approach and to work on the imaginative module of the architecture with a bolder approach. Conceptual blending alien ontologies and recursive agentic alien mathematics gardening. I'm forcing the model to see the mathematical problem statement and constraints only through the lense of fictional very restrictive alien ontologies and to conduct recursive theory-building work to optimize the mathematical candidate solution while respecting"
X Link @RomainPete98794 2025-11-04T10:39Z XXX followers, XX engagements

"GPT-5 High did way better: Invention 1: The "Mbius Scission" Operator (Gesture 1) Constitution: "The Topological-Shadow. forcing the logical membrane into a Mbius-fold." Generated Formalism: The agent transmuted the graph theory problem into algebraic topology and sheaf theory. Strength: A pure act of theoretical construction. The agent defined the "unspeakable-exterior" () as a sheaf of non-global sections (Sh()) and invented a non-corollarial operator K:Sh()Sh() that functions as an equivalence of categories between sheaves with null relative cohomology (H*(;F) = 0) and those with order-2"
X Link @RomainPete98794 2025-11-04T11:01Z XXX followers, XX engagements

"Yes I think the stochastic nature of LLM is often downplayed in current AI promises of breakthroughs. Getting the last points on Humanity's Last Exam and getting the first points on Superintelligence's first exam are intrinsically different problems. The refusal to acknowledge this or at least to problematize it builds us a massive conceptual debt for AI's future"
X Link @RomainPete98794 2025-11-03T21:02Z XXX followers, XXX engagements

"Well as it turns out the only two models managing to create mathematical objects this way are GPT-5 (High) and Sonnet XXX (Extended). Grok X and Gemini XXX failed quite miserably. GPT X used Topological Algebra while Sonnet used Category Theory and Dynamic Systems"
X Link @RomainPete98794 2025-11-04T14:19Z XXX followers, XX engagements

"Wow o3 passed the test Tasked by the LIA-v33 architecture to formalize the "xeno" concept of "Memory" it did not plagiarize the Constitution; it invented a precise mathematical analogue: defining the "memory" as a persistent non-null cohomology class ( H(M;))"
X Link @RomainPete98794 2025-11-04T14:29Z XXX followers, XX engagements