#  @leonpalafox Leon Palafox Leon Palafox posts on X about ai, hay, aos, cuando the most. They currently have [-----] followers and [---] posts still getting attention that total [------] engagements in the last [--] hours. ### Engagements: [------] [#](/creator/twitter::18408506/interactions)  - [--] Week [-------] +150% - [--] Month [-------] +323% - [--] Months [-------] +2,440% - [--] Year [-------] +97% ### Mentions: [--] [#](/creator/twitter::18408506/posts_active)  - [--] Months [--] +6.10% - [--] Year [--] -74% ### Followers: [-----] [#](/creator/twitter::18408506/followers)  - [--] Week [-----] +1.40% - [--] Month [-----] +5.80% - [--] Months [-----] +75% - [--] Year [-----] +108% ### CreatorRank: [-------] [#](/creator/twitter::18408506/influencer_rank)  ### Social Influence **Social category influence** [technology brands](/list/technology-brands) 15.28% [countries](/list/countries) 6.94% [stocks](/list/stocks) 5.56% [currencies](/list/currencies) 4.17% [finance](/list/finance) 3.47% [social networks](/list/social-networks) 3.47% [automotive brands](/list/automotive-brands) 2.08% [celebrities](/list/celebrities) 1.39% [cryptocurrencies](/list/cryptocurrencies) 0.69% [gaming](/list/gaming) 0.69% **Social topic influence** [ai](/topic/ai) 4.86%, [hay](/topic/hay) 4.17%, [aos](/topic/aos) 3.47%, [cuando](/topic/cuando) 3.47%, [gente](/topic/gente) 3.47%, [cmo](/topic/cmo) 2.78%, [vibe coding](/topic/vibe-coding) 2.78%, [eso](/topic/eso) 2.78%, [$googl](/topic/$googl) #1484, [anthropic](/topic/anthropic) 2.78% **Top accounts mentioned or mentioned by** [@osvaldoags](/creator/undefined) [@billymannino](/creator/undefined) [@danielmac8](/creator/undefined) [@jordineil](/creator/undefined) [@zocojonn](/creator/undefined) [@jesusvzr](/creator/undefined) [@deanmckee757](/creator/undefined) [@wordgrammer](/creator/undefined) [@catherder1685](/creator/undefined) [@thetransitguy](/creator/undefined) [@grok](/creator/undefined) [@jessicobra](/creator/undefined) [@noseofdeath](/creator/undefined) [@pfau](/creator/undefined) [@danjeffries1](/creator/undefined) [@scaling01](/creator/undefined) [@luisjacob_gg](/creator/undefined) [@mxquiquebm](/creator/undefined) [@sammoga123](/creator/undefined) [@claro_vi33804](/creator/undefined) **Top assets mentioned** [Alphabet Inc Class A (GOOGL)](/topic/$googl) [Microsoft Corp. (MSFT)](/topic/microsoft) [GrokCoin (GROKCOIN)](/topic/grok) [Tesla, Inc. (TSLA)](/topic/tesla) [Solana (SOL)](/topic/$sol) ### Top Social Posts Top posts by engagements in the last [--] hours "@jack_ceroni Learn to use AI do not shy away from it" [X Link](https://x.com/leonpalafox/status/2022739404184264894) 2026-02-14T18:27Z [----] followers, [---] engagements "Ok no se si es meme pero esto parece mas apocaliptico que nada mas JUST IN: NYC bar opens for people with "AI" partners dining tables optimized to dine with your AI companion. JUST IN: NYC bar opens for people with "AI" partners dining tables optimized to dine with your AI companion" [X Link](https://x.com/leonpalafox/status/2022739652747420076) 2026-02-14T18:28Z [----] followers, [---] engagements "Cmo empezar con vibe coding en ciencia de datos Paso [--] Empieza con un proyecto que domines. Conoces el modelo sus sesgos cmo se optimiza y entiendes bien los datos. Para qu Para poder validar lo que te devuelve el agente. Si algo no cuadra o sale mal lo detectas de inmediato. Paso [--] Ahora s deja que el agente modifique o agregue cosas pero siempre dentro del alcance de tu entendimiento. Aqu aprovechas que escribe cdigo mucho ms rpido que t. Eso s: pdele siempre grficas mtricas y seales que te permitan validar lo que est haciendo. Paso [--] Con un baseline slido deja que el agente se ponga" [X Link](https://x.com/leonpalafox/status/2018711945692516695) 2026-02-03T15:43Z [----] followers, [----] engagements "Y Jsto regresa a operar con nuevos rostros y con nuevo dinero. Igual que lo coment hace algunos aos el modelo de negocio necesita cambiar si en realidad quieren seguir vivos. En retail la divisin que histricamente hace menos dinero es la de perecederos (en Walmart nadie espera hacer dinero con perecederos) y es justo el nicho de Justo. Encima de eso tienen la ltima milla de perecederos histricamente la logstica ms cara. Veamos que sorpresa nos traen en mi opinin necesitan modelos muy precisos de forecast e investigacin de operaciones para tratar de ser rentables pero ya veremos" [X Link](https://x.com/leonpalafox/status/2019021859824750724) 2026-02-04T12:14Z [----] followers, 13.1K engagements "22 aos despus nadie nunca me ha enseado una visualizacin de R que no se pueda hacer con Python. Y ahora con VibeCoding menos" [X Link](https://x.com/leonpalafox/status/2019167771050082362) 2026-02-04T21:54Z [----] followers, 34.4K engagements "Si viste los resultados de Microsoft y Google es fcil quedarse con el crecimiento (Google Cloud al 48% Azure al 39%). Impresionante s. Pero no es lo ms importante. La historia real es la chequera. Estamos viendo cmo se construye el foso econmico ms caro que hemos visto. El CapEx ya no es alto es brutal: Microsoft meti USD [----] mil millones en un solo trimestre y Alphabet est hablando de USD [------] mil millones para [----]. Eso es ms de lo que produce un pas entero en un ao. La guerra del cloud ya no va de software ni de ventas. Va de quin puede controlar el silicio y la capacidad de cmputo. En" [X Link](https://x.com/leonpalafox/status/2019228671199047998) 2026-02-05T01:56Z [----] followers, [----] engagements "Eres junior data scientist y sientes que el vibe coding te est quitando el trabajo No ests solo. Es una realidad: muchas empresas estn contratando menos juniors y en cambio estn dando mejores herramientas a desarrolladores ms senior. Entonces qu toca hacer Volverte ms indispensable que el cdigo que generas. No es solo hacer una query o correr un anlisis es entender lo que ests construyendo y por qu importa. Si tu valor es escribir una query y entregar un dashboard s: eres fcil de reemplazar. Pero si tu valor es hablar con el negocio coordinar equipos mover engranes y llevar conversaciones con" [X Link](https://x.com/leonpalafox/status/2019385444027203799) 2026-02-05T12:19Z [----] followers, 20.7K engagements "Con todo el tema del vibe coding creciendo usar lenguaje natural para construir software Anthropic lanza Claude Opus [---] y se nota el salto. Mejor manejo de contexto largo ms estabilidad en tareas complejas y una experiencia mucho ms fluida para acompaar flujos reales de trabajo. Se siente ms slido cuando lo usas varias horas seguidas y no solo para generar snippets. Cada vez queda ms claro dnde est el valor: tener claridad sobre qu ests construyendo cmo se conecta con el sistema y cmo validar resultados. El reto pasa por el criterio y la comprensin del problema." [X Link](https://x.com/leonpalafox/status/2019568080641565110) 2026-02-06T00:25Z [----] followers, [----] engagements "Alguien por ac me peleaba que no se podra construir un Kernel desde cero con AI Anthropic had [--] AI agents build a C compiler from scratch. 100k lines compiles the Linux kernel $20k [--] weeks. To put that in perspective GCC took thousands of engineers over [--] years to build. (Granted from [----] - however) One researcher and [--] AI agents just built a https://t.co/pk0PxsdTKi Anthropic had [--] AI agents build a C compiler from scratch. 100k lines compiles the Linux kernel $20k [--] weeks. To put that in perspective GCC took thousands of engineers over [--] years to build. (Granted from [----] - however)" [X Link](https://x.com/leonpalafox/status/2019596539589722462) 2026-02-06T02:18Z [----] followers, 20.8K engagements "Y el que faltaba present ya tenemos resultados de Amazon Web Services (AWS) Google y Microsoft . Es sin lugar a duda una de las inversiones privadas ms importantes desde la poca de los ferrocarriles. Y como siempre la nube es un bien pblico" [X Link](https://x.com/leonpalafox/status/2019902487319249177) 2026-02-06T22:34Z [----] followers, [---] engagements "Yo tuve y conoc varios profesores que se vanagloriaban de reprobar alumnos o de lo difcil que era pasar su clase. Aos despus sigo convencido de lo mismo: ese tipo de personas no deberan ser maestros. Deberan dedicarse a otra cosa menos a formar a nuevas generaciones. Porque su objetivo nunca fue educar sino generar frustracin. Tu trabajo no es reprobar gente con exmenes para los que no los preparaste. Tu trabajo es asegurarte de que la mayora de tus alumnos entiende el material que impartes. Presumir que pocos pasan tu clase no es seal de excelencia acadmica. Es casi siempre una forma" [X Link](https://x.com/leonpalafox/status/2020833598836822263) 2026-02-09T12:14Z [----] followers, 23.7K engagements "Esta grfica lo nico que muestra es que se est volviendo a la tranquilidad despus de que durante COVID te daban un puesto con que supieras hacer un import" [X Link](https://x.com/leonpalafox/status/2020933367416660014) 2026-02-09T18:50Z [----] followers, [----] engagements "@data_complex No es verdad el esfuerzo de un modelo mayoritario en espaol no es trivial y no se resuelve con un fine tunning" [X Link](https://x.com/leonpalafox/status/2020986266570063909) 2026-02-09T22:20Z [----] followers, [---] engagements "@daniel_mac8 Dude be careful Gary threatens litigation at the smallest slight" [X Link](https://x.com/leonpalafox/status/2021289432498802780) 2026-02-10T18:25Z [----] followers, [---] engagements "Esto estimados es lo que en verdad puede adelantar a la humanidad varios aos Pharma is COOKED Isomorphic Labs just revealed IsoDDE: an AI system that designs drugs on a computer faster than any pharma R&D doubles AlphaFold [--] on hard targets 20x better than Boltz-2 on antibodies beats the physics gold standard at binding found drug pockets from https://t.co/iQ3xKqU9mw Pharma is COOKED Isomorphic Labs just revealed IsoDDE: an AI system that designs drugs on a computer faster than any pharma R&D doubles AlphaFold [--] on hard targets 20x better than Boltz-2 on antibodies beats the physics gold" [X Link](https://x.com/leonpalafox/status/2021293250292875434) 2026-02-10T18:40Z [----] followers, [----] engagements "Existe mucha gente que la IA es una especie de base de datos donde en el momento que le preguntas algo va y trae el resultado. La realidad es muy distinta si funcionara as le llamaramos base de datos qu padre llamarle "pensar" a encontrar la misma tesis de la que hablas en una base de datos scrappeada sin permiso de nadie y regurgitarte el resultado que a ti te cost meses de trabajo qu padre llamarle "pensar" a encontrar la misma tesis de la que hablas en una base de datos scrappeada sin permiso de nadie y regurgitarte el resultado que a ti te cost meses de trabajo" [X Link](https://x.com/leonpalafox/status/2021583182051606910) 2026-02-11T13:52Z [----] followers, [----] engagements "Pens que ya habamos superado esta discusin pero parece que vale la pena insistir: La IA no es una base de datos. No funciona como un repositorio al que le metes un libro y luego simplemente lo escupe completo cuando alguien lo pide. De hecho es extremadamente difcil que reproduzca citas largas de forma exacta incluso si el texto form parte de su entrenamiento. Lo que hace es distinto: aprende patrones estadsticos del lenguaje y de las estructuras matemticas. No memoriza respuestas como si fueran archivos; construye salidas a partir de representaciones internas. Y esto no es teora. Ya existen" [X Link](https://x.com/leonpalafox/status/2021594737925317084) 2026-02-11T14:38Z [----] followers, 222.6K engagements "La gente ignora vehementemente que el problema no fue que sus inversiones fueran buenas o malas sino que lo estaba haciendo con dinero de otras personas que no le aprobaron hacer eso. Next time you feel bad for missing something that mooned Imagine being SBF: - Put $500m into Anthropic which would now be worth $70b. - Held $60m in $SOL at $8 that would've been worth $2.1b at the top. - Put $100m into Mysten Labs that would now be worth over $800m. - https://t.co/6axraF18DC Next time you feel bad for missing something that mooned Imagine being SBF: - Put $500m into Anthropic which would now be" [X Link](https://x.com/leonpalafox/status/2021597427250119034) 2026-02-11T14:49Z [----] followers, [----] engagements "@JordiNeil Es bien sabido que hay una paradoja en Deep Learning donde es muy bueno generalizando aun cuando se le entren para hacer overfitting https://lis.csail.mit.edu/pubs/kawaguchi-techreport18.pdf https://lis.csail.mit.edu/pubs/kawaguchi-techreport18.pdf" [X Link](https://x.com/leonpalafox/status/2021600759960187217) 2026-02-11T15:02Z [----] followers, [----] engagements "Las respuestas a este post me indican que si que debemos segur discutiendo esto. La poblacin realmente cree que sabe de tecnologa cuando no tiene idea de la diferencia entre ; - Motor de Bsqueda - Base de Datos - LLM Creen que se pueden usar como palabras intercambiables y no lo son Pens que ya habamos superado esta discusin pero parece que vale la pena insistir: La IA no es una base de datos. No funciona como un repositorio al que le metes un libro y luego simplemente lo escupe completo cuando alguien lo pide. De hecho es extremadamente difcil Pens que ya habamos superado esta discusin pero" [X Link](https://x.com/leonpalafox/status/2021939374020026791) 2026-02-12T13:28Z [----] followers, [----] engagements "Con los avances que he visto no me sorprendera que escribir cdigo muera. Es una tarea sper mecnica donde tenemos que poner capas de abstraccin para poder hacer algo til (si un lenguaje de programacin es una abstraccin para humanos) Con suficiente informacin en binario y hexadecimal podemos crear IAs que puedan crear cualquier cosa escrita directamente en binario y toda interaccin humana terminara. Lo que dudo mucho es que suceda en los prximos [--] aos y menos en este ao Elon Musk thinks coding dies this year. Not evolves. Dies. By December AI wont need programming languages. It generates" [X Link](https://x.com/leonpalafox/status/2021957496592548196) 2026-02-12T14:40Z [----] followers, 13.5K engagements "@claro_vi33804 Fabuloso tu poder de discusin" [X Link](https://x.com/leonpalafox/status/2022041586671276358) 2026-02-12T20:14Z [----] followers, [--] engagements "Machine Learning sucks. La frase es de Yann LeCun. Y aunque suena exagerada tiene fondo. Nuestros modelos actuales son impresionantes pero profundamente ineficientes. Aprenden con ocanos de datos lo que la inteligencia biolgica capta con unas cuantas experiencias. En Duke LeCun insisti en algo incmodo: ms datos no es igual a AGI. Los LLMs autoregresivos predicen token por token; mientras ms larga la respuesta mayor la probabilidad de desviarse. Escalan s. Pero no necesariamente razonan. Por eso su empuje hacia JEPA y world models no es capricho. Es un cambio de direccin: menos obsesin por" [X Link](https://x.com/leonpalafox/status/2022067210525495548) 2026-02-12T21:56Z [----] followers, [----] engagements "Yo he estado a punto de pagar los [---] dlares al mes ms de una vez ever have certain moments where the only thing in your mind is: "what the fuck" i mean you gotta admire the narrative discipline the execution & the go to market. this is incredible. https://t.co/Adh8jpXiBq ever have certain moments where the only thing in your mind is: "what the fuck" i mean you gotta admire the narrative discipline the execution & the go to market. this is incredible. https://t.co/Adh8jpXiBq" [X Link](https://x.com/leonpalafox/status/2022109207441653887) 2026-02-13T00:42Z [----] followers, [----] engagements "En nuestro episodio de ayer de la clase de Deep Learning: Hay una pregunta muy sencilla que suele revelar si alguien realmente entiende cmo funcionan los modelos de lenguaje: Puede explicar la diferencia entre un token y un embedding Un token es la unidad bsica con la que el modelo procesa texto. Es cmo fragmenta lo que le damos y lo que genera. Ah se miden costos lmites de contexto y velocidad. Un embedding es la representacin numrica del significado. Es el momento en que el lenguaje se convierte en un vector dentro de un espacio matemtico donde las relaciones semnticas se pueden medir." [X Link](https://x.com/leonpalafox/status/2022361975985181124) 2026-02-13T17:27Z [----] followers, 14.6K engagements "@zocojonn No hay mucho ingeniero de IA que sabe eso pero no sabe exactamente cmo se calcula un embedding o qu cosa es gradiente descendiente esos trminos son ms de aplicaciones" [X Link](https://x.com/leonpalafox/status/2022694040748130462) 2026-02-14T15:26Z [----] followers, [--] engagements "La gente cree que esto con ciclovas se arregla pero no lo principal es transporte pblico. Esto es un problema que le debemos a administraciones de izquierda que en [--] aos nos han dado una sola lnea de metro y hasta el metrobs (que es un parche) parece que se les olvid en pos de trolebuses. Brugada sigue presumiendo que va a haber dos lneas nuevas de la cual se ve cero avance La CDMX sigue siendo de las ciudades ms contaminadas del continente y con mayor trfico EN EL MUNDO y esto no es casualidad. Y todo a causa de sus polticas de abandonar la expansin del metro sustituirlo por autobuses y" [X Link](https://x.com/leonpalafox/status/2022714165031219241) 2026-02-14T16:46Z [----] followers, [----] engagements "@OsvaldoAGS Te invito a ver cuantas ciclovas hay en Tokio u Osaka" [X Link](https://x.com/leonpalafox/status/2022718646326563108) 2026-02-14T17:04Z [----] followers, [---] engagements "@moaimx Que modelo estas usando La mayora de los foros dicen que Claude no tiene mayor tema" [X Link](https://x.com/leonpalafox/status/2022740357008756828) 2026-02-14T18:30Z [----] followers, [--] engagements "Existe una clara estrategia para hacer frente a la IA Estamos entrando a una nueva divisin social frente a la inteligencia artificial. No es solo quin la usa y quin no. La diferencia real est en la combinacin entre mentalidad y capacidad. Hoy conviven cuatro perfiles: Quienes quieren aprender pero no tienen acceso a herramientas. Quienes tienen acceso pero la desestiman. Quienes ni tienen acceso ni inters. Quienes quieren aprender y adems se equipan para usarla estratgicamente. La evidencia sobre adopcin tecnolgica muestra que la ventaja no proviene solo del acceso ni solo de la intencin." [X Link](https://x.com/leonpalafox/status/2022744284462727182) 2026-02-14T18:46Z [----] followers, [----] engagements "@OsvaldoAGS @JesusVZR Me podras por favor compartir el link donde muestranq ue mueven ms gente para el caso de la CDMX (y no una ciudad Europea con muchos menos habitantes)" [X Link](https://x.com/leonpalafox/status/2022744574721167580) 2026-02-14T18:47Z [----] followers, [--] engagements "Anthropic hitting a $14B run rate in just three years is incredible to see. The real signal here is Claude Code: its already at a $2.5B run rate and driving 4% of all GitHub commits. We are watching the developer landscape shift in real-time" [X Link](https://x.com/leonpalafox/status/2022110512969064735) 2026-02-13T00:48Z [----] followers, [---] engagements "Esto es interesante y muy real. Asi como hoy hay personas que "pues hago lo que dice en la maquina" y otras que "hacen lo que ponen en la maquina" In [--] years there will be two classes of people. Economists call it the "K-shaped economy" - and the next 2-3 years will decide which line you're on. An overclass that uses AI as a lever to build wealth automate income and make decisions at a speed no human can compete https://t.co/pN8QZBz9cv In [--] years there will be two classes of people. Economists call it the "K-shaped economy" - and the next 2-3 years will decide which line you're on. An" [X Link](https://x.com/leonpalafox/status/2022741876575736098) 2026-02-14T18:36Z [----] followers, [----] engagements "La gran realidad es que el 80% de las cosas que se hagan con IA sern triviales The punchline here is incredible https://t.co/qoznql8t4X The punchline here is incredible https://t.co/qoznql8t4X" [X Link](https://x.com/leonpalafox/status/2022746673810743509) 2026-02-14T18:56Z [----] followers, [----] engagements "Madre. [----] varos. podra probarlo un mes u ver que onda. The upgraded Deep Think mode is rolling out now in the @GeminiApp for Google AI Ultra subscribers. For scientific researchers and developers were opening a Vertex AI Early Access Program for the API. Start discovering https://t.co/8OhnV65pW5 https://t.co/WGSlntIFLc The upgraded Deep Think mode is rolling out now in the @GeminiApp for Google AI Ultra subscribers. For scientific researchers and developers were opening a Vertex AI Early Access Program for the API. Start discovering https://t.co/8OhnV65pW5 https://t.co/WGSlntIFLc" [X Link](https://x.com/leonpalafox/status/2022798080244924492) 2026-02-14T22:20Z [----] followers, 18.7K engagements "Me causa curiosidad saber que fue lo que necesit el Pentgono particularmente de la IA de @AnthropicAI . Uno pensara que tienen suficiente informacin" [X Link](https://x.com/leonpalafox/status/2022914597750190444) 2026-02-15T06:03Z [----] followers, [----] engagements "@Prazkat Dispararle a los gays en Los Feliz es una clara muestra de lo Woke que es" [X Link](https://x.com/anyuser/status/1732107722303136044) 2023-12-05T18:40Z [----] followers, 59.7K engagements "La IA puede apoyar a descubrir nuevas cosas en la fisica. Ciertamente te puede ayudar a hacer la tarea GPT-5.2 derived a new result in theoretical physics. Were releasing the result in a preprint with researchers from @the_IAS @VanderbiltU @Cambridge_Uni and @Harvard. It shows that a gluon interaction many physicists expected would not occur can arise under specific GPT-5.2 derived a new result in theoretical physics. Were releasing the result in a preprint with researchers from @the_IAS @VanderbiltU @Cambridge_Uni and @Harvard. It shows that a gluon interaction many physicists expected would" [X Link](https://x.com/leonpalafox/status/2022487887112212847) 2026-02-14T01:47Z [----] followers, [----] engagements "@OsvaldoAGS Mas del doble jajajaja" [X Link](https://x.com/leonpalafox/status/2022733444753821744) 2026-02-14T18:03Z [----] followers, [--] engagements "@jmtoralc Lo ms pro es hacerlo en wortrees. Yo lo hice una vez y se acab me dinero en menos de una hr" [X Link](https://x.com/leonpalafox/status/2022795285877973138) 2026-02-14T22:09Z [----] followers, [---] engagements "@fernetcontonica @futboltrotters En Mxico la mam de Oliver (nada de Tsubasa) le agradece a la virgen de Guadalupe" [X Link](https://x.com/leonpalafox/status/2022796204736729243) 2026-02-14T22:12Z [----] followers, [---] engagements "Para ser honestos me gustara poder ir en tren al trabajo tambin Solo puedo decir que amo vivir en Toluca por el simple hecho de poder utilizar un tren para ir a la CDMX https://t.co/dFCxNI0WNC Solo puedo decir que amo vivir en Toluca por el simple hecho de poder utilizar un tren para ir a la CDMX https://t.co/dFCxNI0WNC" [X Link](https://x.com/leonpalafox/status/2021287302233034779) 2026-02-10T18:17Z [----] followers, 10.5K engagements "Para m la mejor prueba del cido que alguien sabe o no de IA es que en verdad sabe la diferencia entre un Token y un Embedding" [X Link](https://x.com/leonpalafox/status/2022331409592189351) 2026-02-13T15:25Z [----] followers, 42.9K engagements "Yo sigo mucho a Ray por LinkedIn y he ledo sus libros. Me parece una buena lectura aunque exagera en algunas cosas https://t.co/tjmbT5ytUN https://t.co/tjmbT5ytUN" [X Link](https://x.com/leonpalafox/status/2023185545283940370) 2026-02-15T23:59Z [----] followers, [----] engagements "You can generate multiple parallel answers and start auto pruning the tree which is probably what they did for o3. Yann is talking about a single autoregressive run and he is right on that one the more tokens you generate the more likely it is to pick a low probability token and start going astray" [X Link](https://x.com/leonpalafox/status/1871199526725267583) 2024-12-23T14:21Z [----] followers, [---] engagements "@deanmckee757 @wordgrammer What If you work in a bank you get tabular data as daily transactions. If you work at retail sales and client behavior also comes as tabular data" [X Link](https://x.com/leonpalafox/status/1871638949627932908) 2024-12-24T19:27Z [----] followers, [--] engagements "People saying that GenAI can replace traditional ML are my favorite people right away you know theyre full of bull" [X Link](https://x.com/leonpalafox/status/1871639243048784000) 2024-12-24T19:29Z [----] followers, [---] engagements "@deanmckee757 @wordgrammer And why would you spend that much compute to solve marginally better something that XGBoost can solve Doesnt make an iota of sense I used to score millions of clients each week. Saying that is better to use GenAI instead does not make much sense" [X Link](https://x.com/leonpalafox/status/1871645440334299539) 2024-12-24T19:53Z [----] followers, [--] engagements "@Lons Just tell them is the sequel of Brad Pitt movie" [X Link](https://x.com/leonpalafox/status/1871759990442320127) 2024-12-25T03:28Z [----] followers, [--] engagements "Im sure those kids will remember their folks grinding to make Grok tell slightly less wrong answers and not the good time they could have spent with their parents. At @xai we were planning to grind through Christmas. But then @ibab had a strange experience with three apparitions 👻👻👻. The regular grinding will continue after tomorrow. At @xai we were planning to grind through Christmas. But then @ibab had a strange experience with three apparitions 👻👻👻. The regular grinding will continue after tomorrow" [X Link](https://x.com/leonpalafox/status/1871818791975629011) 2024-12-25T07:22Z [----] followers, [---] engagements "@bryan_johnson I mean 90% of people still dont understand the Internet" [X Link](https://x.com/leonpalafox/status/1871952257257701444) 2024-12-25T16:12Z [----] followers, [--] engagements "@billymannino Even at that Apple is also digital which means there is a quantization involved which means theres a loss. Unless is bulbs all the way from recording to output there are losses" [X Link](https://x.com/leonpalafox/status/1872406828216402395) 2024-12-26T22:19Z [----] followers, 22K engagements "@Catherder1685 @billymannino Yes but thats not the point the point is that there isnt a truly 100% fidelity recording" [X Link](https://x.com/leonpalafox/status/1872515384106209738) 2024-12-27T05:30Z [----] followers, [---] engagements "@OCopium @billymannino Speaker has a piezoelectric piece that transforms current into vibrations the speaker in and of itself offers way more fidelity than the AC/DC Also being driven by electronics doesnt imply loss of fidelity as you can have analog electronics. And not all electronics are digital" [X Link](https://x.com/leonpalafox/status/1872656288351408621) 2024-12-27T14:50Z [----] followers, [---] engagements "@Catherder1685 @billymannino Not really since my first post Ive been saying that quantization in a digital device implies that there is no 100% fidelity" [X Link](https://x.com/leonpalafox/status/1872656651049992310) 2024-12-27T14:51Z [----] followers, [--] engagements "@StevenPulteFam Why spend 4k in your 20s for a watch Put it in the SP500 and buy a car in your 40s" [X Link](https://x.com/leonpalafox/status/1873186901953527826) 2024-12-29T01:58Z [----] followers, [---] engagements "@the_transit_guy lol comparing Spain (a whole country) with Houston a single city" [X Link](https://x.com/leonpalafox/status/1873535343443800434) 2024-12-30T01:03Z [----] followers, [---] engagements "@nkalamb @nickjanetos @Rider_Nytimus @_Zeets I have a hard time thinking that a ton of very educated scholars got the translations wrong but she got it right" [X Link](https://x.com/leonpalafox/status/1873739410300387499) 2024-12-30T14:34Z [----] followers, [----] engagements "@heartsalty I dont think Fagles or Lattimore were just bouncing from other translations" [X Link](https://x.com/leonpalafox/status/1873746506232041477) 2024-12-30T15:02Z [----] followers, [---] engagements "Image by @grok" [X Link](https://x.com/leonpalafox/status/1873875140200816704) 2024-12-30T23:33Z [----] followers, [--] engagements "@HanchungLee In every job I've had we had a team of DSs and a team of DEs so one would experiment in Jupyter/Rstudio and the opther would create the pipelines is Scala/Spark/Databricks" [X Link](https://x.com/leonpalafox/status/1874108437581492272) 2024-12-31T15:00Z [----] followers, [---] engagements "@SyncereX8 @kingofbrethren What's wron with Brad Pitt as a Mexican" [X Link](https://x.com/leonpalafox/status/1874121679179571326) 2024-12-31T15:53Z [----] followers, [----] engagements "@lemire Go further if you had invested [--] USD in Intel in the [----] you would have exactly the same amount of money and if you had done it in [----] you've lost [--] times your money" [X Link](https://x.com/leonpalafox/status/1874207817651159357) 2024-12-31T21:35Z [----] followers, [---] engagements "The people wondering why Grok [--] hasn't shipped.did you know what happened when Musk promised Teslas for a certain day Musk is terrible at reaching deadlines because he overpromises" [X Link](https://x.com/leonpalafox/status/1875200012092850545) 2025-01-03T15:18Z [----] followers, [---] engagements "@sergiooliveiram Nissan las Peugeot se les funde un faro y se les prenden mil testigos" [X Link](https://x.com/leonpalafox/status/1875572397052719407) 2025-01-04T15:58Z [----] followers, [--] engagements "@alexrmoskowitz The only function of economic forecasting is to make astrology look respectable" [X Link](https://x.com/leonpalafox/status/1876054048937476252) 2025-01-05T23:51Z [----] followers, [----] engagements "@GoodPoliticGuy Noup it was the overall willingness of the Dems to not look the rural bit and focus instead on the college educated" [X Link](https://x.com/leonpalafox/status/1876065812550987873) 2025-01-06T00:38Z [----] followers, [----] engagements "@Coolio_Art Those arent accurate raptors they are way too big" [X Link](https://x.com/leonpalafox/status/1876132648097370256) 2025-01-06T05:04Z [----] followers, [---] engagements "@Beach_Wolf69 @WomanDefiner I think he makes a better service exposing these kind of spenders. To show that even people who win [--] figures (in this case a couple) can blow all their money and have a hard time getting out of debt" [X Link](https://x.com/leonpalafox/status/1876644333052473741) 2025-01-07T14:57Z [----] followers, [----] engagements "@DarkwebShaman In Japan they don't have black people and they still have those doors" [X Link](https://x.com/leonpalafox/status/1876997946882859506) 2025-01-08T14:22Z [----] followers, [--] engagements "@jessicobra How is Capitalism bad" [X Link](https://x.com/leonpalafox/status/1876999047426277870) 2025-01-08T14:27Z [----] followers, [----] engagements "@hercine3 @jessicobra No I'm saying that the acumulaton of wealth is not a feature of capitalism is a feature of human nature. Thus your one complain about capitalism is not exclusive of it" [X Link](https://x.com/leonpalafox/status/1877026203535634867) 2025-01-08T16:14Z [----] followers, [---] engagements "@ajordannafa You cant use SGd if the matrix X is millions of element large computationally speaking is way more effective to use SGD" [X Link](https://x.com/leonpalafox/status/1877217194246803727) 2025-01-09T04:53Z [----] followers, [---] engagements "@nhfoley @bryancsk All of Mexico City is concrete buildings in a VERY earthquake prone zone" [X Link](https://x.com/leonpalafox/status/1878312906447614191) 2025-01-12T05:27Z [----] followers, [--] engagements "I love AI Data Teams where there isn't a single person that really knows about AI" [X Link](https://x.com/leonpalafox/status/1878880998865420438) 2025-01-13T19:05Z [----] followers, [---] engagements "@GayBearRes In many companies this is the case. Im seeing it as well. Most of our consultants (Accenture Indigene etc) have their tech teams based in India" [X Link](https://x.com/leonpalafox/status/1879213399802007906) 2025-01-14T17:06Z [----] followers, [--] engagements "And maybe she is being paid by TikTok to say this. This video by popular TikTok personality Soupy is going viral with [---] million likes and [---] million views in [--] hours. Gen z is not happy. https://t.co/bhfCPv1do7 This video by popular TikTok personality Soupy is going viral with [---] million likes and [---] million views in [--] hours. Gen z is not happy. https://t.co/bhfCPv1do7" [X Link](https://x.com/leonpalafox/status/1879214681098998052) 2025-01-14T17:11Z [----] followers, [---] engagements "Theres a big difference Bad Bunny wants to cater to an American audience but youre not seeing Fallon in Telemundo Bad Bunny telling Jimmy Fallon how he's improved his English over the years and that it's Fallon's turn to improve his Spanish is EXACTLY what these gringos need to hear. They always expect to be catered to and get uncomfortable when they aren't. Bad Bunny telling Jimmy Fallon how he's improved his English over the years and that it's Fallon's turn to improve his Spanish is EXACTLY what these gringos need to hear. They always expect to be catered to and get uncomfortable when they" [X Link](https://x.com/leonpalafox/status/1879305476804059336) 2025-01-14T23:11Z [----] followers, [---] engagements "@kopeboy @pitdesi The Disney parks revenue is about [--] billion USD a year so .not much" [X Link](https://x.com/leonpalafox/status/1879858411204137309) 2025-01-16T11:49Z [----] followers, [--] engagements "@LeeWinterOz @pot8um Because it was hard from the developer POV few people remember the hell that was the yearly release cycle now you have incremental upgrades" [X Link](https://x.com/leonpalafox/status/1879882076981436696) 2025-01-16T13:23Z [----] followers, [---] engagements "@ZebitasMartinex Convenientemente te saltaste VirtualBoy N64 y Game Cube" [X Link](https://x.com/leonpalafox/status/1880110586505363788) 2025-01-17T04:31Z [----] followers, [---] engagements "@445543636536s @the_transit_guy Also they are comparing a 450million continent with a [---] million country Just by virtue of that is obvious main cities will have more people Go and compare with India or China" [X Link](https://x.com/leonpalafox/status/1880613422393635026) 2025-01-18T13:49Z [----] followers, [--] engagements "Since the Wii every time Nintendo releases a new console haters talk about graphics. People dont realize that Nintendo already went the path of latest graphics chipset and it failed miserably. The N64 was more powerful than the PS One and the Cube was more powerful than the PS2" [X Link](https://x.com/leonpalafox/status/1881185236169498741) 2025-01-20T03:41Z [----] followers, [----] engagements "@Logan96194157 @NoseOfDeath The Cube used CDs" [X Link](https://x.com/leonpalafox/status/1881304098752040976) 2025-01-20T11:33Z [----] followers, [---] engagements "Sama is pretty much Musk [---] he has been hyping GenAI for the past month and now he is like uhhh no no that is justo Twitter being Twitter twitter hype is out of control again. we are not gonna deploy AGI next month nor have we built it. we have some very cool stuff for you but pls chill and cut your expectations 100x twitter hype is out of control again. we are not gonna deploy AGI next month nor have we built it. we have some very cool stuff for you but pls chill and cut your expectations 100x" [X Link](https://x.com/leonpalafox/status/1881320870989406307) 2025-01-20T12:40Z [----] followers, [---] engagements "@GdGoncalo @NoseOfDeath Yeah no one argues that the storage medium in PS1 allowed for CG videos which the N64 sorely lacked but game rendering was superior in the N64" [X Link](https://x.com/leonpalafox/status/1881415086096752869) 2025-01-20T18:54Z [----] followers, [--] engagements "@B2RKN @spectatorindex All Asian countries have very aggressive laws against drugs. Singapore one of the most advanced cities has the death penalty for possession of many drugs" [X Link](https://x.com/leonpalafox/status/1881527786927714310) 2025-01-21T02:22Z [----] followers, [---] engagements "Tan fcil que era para Musk no hacer el saludo y ya Found this in Reddit seems like it might be useful for some people. https://t.co/Mg6tCg6aAB Found this in Reddit seems like it might be useful for some people. https://t.co/Mg6tCg6aAB" [X Link](https://x.com/leonpalafox/status/1882487210374807566) 2025-01-23T17:55Z [----] followers, [---] engagements "@__Siffa @filippie509 3% Llama is estimated to have been trained on 16K processors also a ton of the hardware goes towards inference not just training" [X Link](https://x.com/leonpalafox/status/1883991674222961016) 2025-01-27T21:33Z [----] followers, [--] engagements "@quesadillino @YosoyPedrero Fridays quebr" [X Link](https://x.com/leonpalafox/status/1886083176050708536) 2025-02-02T16:04Z [----] followers, [---] engagements "Yet another GPT wrapper that companies want to pass as an innovative feature whyshould I pay you [--] USD for something I can do in my GPT account" [X Link](https://x.com/leonpalafox/status/1886819169297162649) 2025-02-04T16:48Z [----] followers, [---] engagements "Esto pasa para todo no solo con atletas. Dios est ah para nuestras victorias pero cuando pasa una desgracia jams lo culpamos Curious that talented athletes frequently credit God when they win but we rarely see them blame God when they lose. Curious that talented athletes frequently credit God when they win but we rarely see them blame God when they lose" [X Link](https://x.com/leonpalafox/status/1889052803332448665) 2025-02-10T20:44Z [----] followers, [---] engagements "This week in class we explored how LLMs are becoming the foundation for enterprise AI apps and what tools and concepts are powering that shift. Heres the architecture we discussed: 🔗 Taxonomies help structure knowledge and bring consistency to LLM interactions essential for governance 📚 RAG (Retrieval-Augmented Generation) connects LLMs to external knowledge sources to improve accuracy and reduce hallucination 🤖 Agents allow models to take actions use tools and handle multi-step tasks 🧩 Model Context Protocol (MCP) a new standard from Anthropic that lets multiple applications and tools" [X Link](https://x.com/leonpalafox/status/1903079049200799872) 2025-03-21T13:39Z [----] followers, [---] engagements "@ostrisai Because it relates to the user and avoids paying a fee to do it or adding [--] thousand selfies. Same reason things are popular" [X Link](https://x.com/leonpalafox/status/1905634526430102000) 2025-03-28T14:54Z [----] followers, [--] engagements "@paulg @mattyglesias Any random Chinese city looks better than NYC services public transport everything" [X Link](https://x.com/leonpalafox/status/1906731292500082823) 2025-03-31T15:32Z [----] followers, [----] engagements "@skirano RAG was never about context was about reducing hallucinations. Long context does nothing to do that" [X Link](https://x.com/leonpalafox/status/1908651786115625166) 2025-04-05T22:43Z [----] followers, [----] engagements "Madagascar.twice Delete one franchise FOREVER https://t.co/ojBZeIUOfE Delete one franchise FOREVER https://t.co/ojBZeIUOfE" [X Link](https://x.com/leonpalafox/status/1910427172424343899) 2025-04-10T20:18Z [----] followers, [---] engagements "@My1xT @atom_jarvis @BridgetCambria8 @FrankLuntz Most CRM do the same thing they warm up the mails because if GMAIL detects a massive splurge of mails from the same address they will send them to junk mail" [X Link](https://x.com/leonpalafox/status/1911235248001605655) 2025-04-13T01:49Z [----] followers, [---] engagements "@JesseSchoberg @levelsio This is disingenuous AB InBeV operates multiple brands in multiple countries its an amalgamation of many companies" [X Link](https://x.com/leonpalafox/status/1911466510511722892) 2025-04-13T17:08Z [----] followers, [----] engagements ""This order promotes integrating AI into K-12 education by encouraging AI literacy teacher training and partnerships with private and academic institutions. It establishes a White House Task Force on AI Education and directs federal agencies to prioritize AI in discretionary grant programs for teacher training"" [X Link](https://x.com/leonpalafox/status/1916304800880250927) 2025-04-27T01:34Z [----] followers, [--] engagements "@jxmnop The fix is easy do as any other science and accept pretty much every paper and make conferences a networking opportunity" [X Link](https://x.com/leonpalafox/status/1918884293163565149) 2025-05-04T04:24Z [----] followers, [---] engagements "@ICBItsNotBetty @ArmandDoma No I decided to go to Asia to study robotics in a plan thats way better than anything Harvard has" [X Link](https://x.com/leonpalafox/status/1926826894994952483) 2025-05-26T02:25Z [----] followers, [--] engagements "@hecate40 @kareem_carr CEO of Meta - Grandson of immigrants CEO of Tesla - Immigrant CEO of Google - Immigrant CEO of OpenAI - also from a family of immigrants. Man I wonder how can America innovation can get down if you stop preventing immigrants from going to your country" [X Link](https://x.com/leonpalafox/status/1927693022617116895) 2025-05-28T11:46Z [----] followers, [---] engagements "These three legends are basically the north star I follow when it comes to story telling humor and energy" [X Link](https://x.com/leonpalafox/status/1932115487884001605) 2025-06-09T16:40Z [----] followers, [---] engagements "Were quick to criticize ChatGPT for killing our critical thinking. But where was this outrage when we spent years normalizing rote memorization as the default way of teaching Many of us were forced to memorizeby heartthe European countries and their capitals. How much of that knowledge do we really retain today And how many schools still teach math the same way history is taught: by reciting formulas without understanding how or why they work Lets be honest. Long before ChatGPT we already had a school system that often discouraged curiosity punished questioning and prioritized repetition over" [X Link](https://x.com/leonpalafox/status/1935563615228018718) 2025-06-19T05:01Z [----] followers, [---] engagements "@MushtaqBilalPhD Academic Publishers is one of the greates transferrs of cash from Government to private coorporations that we've ever seen. If anyone saw ti from that lens I'm sure governments would stop funding these" [X Link](https://x.com/leonpalafox/status/1936842704958484627) 2025-06-22T17:44Z [----] followers, 12K engagements "@ankurnagpal Zoom tried to compete against Microsoft in their own game and of course no one can challenge MSFT distributionc apabilities. It literally had zero edge against Teams" [X Link](https://x.com/leonpalafox/status/1937666251746095384) 2025-06-25T00:16Z [----] followers, [---] engagements "@jboysen0 @pfau I refer to CS only bros who know shit about wet experiments yet want to change health and medicine" [X Link](https://x.com/leonpalafox/status/1939912472971952177) 2025-07-01T05:02Z [----] followers, [---] engagements "@haro_ca_ Why" [X Link](https://x.com/leonpalafox/status/1942659348595122367) 2025-07-08T18:57Z [----] followers, [--] engagements "@pfau Whats even the point Is like having AI art created to be attended solely by robots" [X Link](https://x.com/leonpalafox/status/1943699913633636649) 2025-07-11T15:52Z [----] followers, [--] engagements "He has a PhD in Machine Learning and needed an expert to tell him using Transformers for tabular data is a bad idea" [X Link](https://x.com/leonpalafox/status/1948022426123772389) 2025-07-23T14:08Z [----] followers, [---] engagements "Someone should make the experiment try fine tuning on all last reviews and then have a cohorts of humans reviewing and an LLM reviewer and see how much acceptance rates change NeurIPS reviews are now complete trash. NeurIPS reviews are now complete trash" [X Link](https://x.com/leonpalafox/status/1948749841330327669) 2025-07-25T14:19Z [----] followers, [---] engagements "@2prime_PKU Hey at least you know he or she didnt use an LLM 😆" [X Link](https://x.com/leonpalafox/status/1948795400883761486) 2025-07-25T17:20Z [----] followers, [---] engagements "Hay dos chief scientists de AI en meta We are excited to announce that @shengjia_zhao will be the Chief Scientist of Meta Superintelligence Labs Shengjia is a brilliant scientist who most recently pioneered a new scaling paradigm in his research. He will lead our scientific direction for our team. Let's go 🚀 https://t.co/D93KQWIvFl We are excited to announce that @shengjia_zhao will be the Chief Scientist of Meta Superintelligence Labs Shengjia is a brilliant scientist who most recently pioneered a new scaling paradigm in his research. He will lead our scientific direction for our team." [X Link](https://x.com/leonpalafox/status/1948899798859612271) 2025-07-26T00:14Z [----] followers, [---] engagements "Dime que no sabes nada de la industria de autos sin decirme que no sabes nada de la industria" [X Link](https://x.com/leonpalafox/status/1949481258737659999) 2025-07-27T14:45Z [----] followers, [--] engagements "Who wants to arrive [--] before to the stations and in [--] you are already in the train. Nooo I need to arrive [--] hours early either lo LAX or SFO (Both are a mess) lineup with strangers and take my shoes off and pc from my bag eat expensive airport food and theeeen get in my plane" [X Link](https://x.com/leonpalafox/status/1950924336333775352) 2025-07-31T14:19Z [----] followers, 12.2K engagements "@MyPronounsRFOff @NiMingDa_888 Please tell me who did Venezuela sold most of its oil to" [X Link](https://x.com/leonpalafox/status/1952595436747010151) 2025-08-05T05:00Z [----] followers, [---] engagements "@LinkofSunshine Average return rates are 10% thats above inflation and that gives you to live comfortably with 80k and leave the money growing until youre [--] to move it to a lower risk fund" [X Link](https://x.com/leonpalafox/status/1952786327243461047) 2025-08-05T17:38Z [----] followers, [----] engagements "@JUSTDEMITOY Shakira stopped being a writer a long time ago most of her hits are pop/rap collaborations a long shot from she first started. She does write from time to time but not like other not established ones" [X Link](https://x.com/leonpalafox/status/1953868839478772185) 2025-08-08T17:20Z [----] followers, [---] engagements "Is this the moment OpenAI jumps the shark" [X Link](https://x.com/leonpalafox/status/1953909727248658742) 2025-08-08T20:02Z [----] followers, [---] engagements "95% of GenAI pilots are failing. Why All the problems weve seen for years in Data Science dirty data unguided models lack of clear business cases and no usage metrics didnt go away with Generative AI. In fact GenAI only amplified these issues. If your traditional ML models couldnt survive in production theres no way a GenAI model will. Most failures are not about the models themselves but about deeper structural problems in organizations. My advice Start with clear measurable business use cases. Define and standardize your prompts and architectures. Track usage and adoption from day one." [X Link](https://x.com/leonpalafox/status/1958619340552405372) 2025-08-21T19:56Z [----] followers, [---] engagements "@dinels12 Ya varios te lo dijeron la IA que usa OpenAI es DeepLearning que es una rama de Machine Learning que en realidad no es ms que un rebranding de Redes Neuronales. Agarra un libro y ve eso" [X Link](https://x.com/leonpalafox/status/1958633012914262039) 2025-08-21T20:51Z [----] followers, [---] engagements "Link: https://cloud.google.com/blog/products/infrastructure/measuring-the-environmental-impact-of-ai-inference https://cloud.google.com/blog/products/infrastructure/measuring-the-environmental-impact-of-ai-inference" [X Link](https://x.com/leonpalafox/status/1958682384150262207) 2025-08-22T00:07Z [----] followers, [---] engagements "@Dan_Jeffries1 These people have no idea how medicine or Law even works. There you have the godfather of AI Hinton saying that radiologists would be without jobs now. These are people that rarely have gotten out to a world outside of their computers" [X Link](https://x.com/leonpalafox/status/1959358336828166291) 2025-08-23T20:53Z [----] followers, [---] engagements "@francoisfleuret Yes I always teach it like in my ML class. We really have snippets of the phenomenon we are trying to model and we have the hubris to think that our model will be able to map that space. Sometimes it works sometimes it doesnt" [X Link](https://x.com/leonpalafox/status/1961525392168423835) 2025-08-29T20:24Z [----] followers, [--] engagements "In every fusion sensor system you need a failsafe most smart cars have radar or sonar for their autostop systems (Nissan BYD etc) because sonar gives you a very fail proof system of estimating distance to an incoming obstacle. Video does not. Theres also this video that shows how you can Willy Coyote a Tesla https://youtu.be/IQJL3htsDyQsi=rGOCPPgCrYEpTQgF https://youtu.be/IQJL3htsDyQsi=rGOCPPgCrYEpTQgF" [X Link](https://x.com/leonpalafox/status/1962026559155405250) 2025-08-31T05:35Z [----] followers, [---] engagements "Quizas ek futuro del internet sea una interaccin masiva con LLMs que son conectados con fuentes fidedignas (periodicos gobiernos) i never took the dead internet theory that seriously but it seems like there are really a lot of LLM-run twitter accounts now i never took the dead internet theory that seriously but it seems like there are really a lot of LLM-run twitter accounts now" [X Link](https://x.com/leonpalafox/status/1963417664530907328) 2025-09-04T01:43Z [----] followers, [---] engagements "@daniel_mac8 OpenAI just discovered what the rest of the community has known for years" [X Link](https://x.com/leonpalafox/status/1964378064588317171) 2025-09-06T17:20Z [----] followers, [---] engagements "@Maav_BR @martinmbauer An LLM is a transformer. The real answer is that they did not use a transformer trained on internet text but on model results" [X Link](https://x.com/leonpalafox/status/1966286629464928310) 2025-09-11T23:43Z [----] followers, [---] engagements "I just published OpenAIs $300 Billion Gamble on Unreliable Answers https://medium.com/p/openais-300-billion-gamble-on-unreliable-answers-3f3959da4bfdsource=social.tw https://medium.com/p/openais-300-billion-gamble-on-unreliable-answers-3f3959da4bfdsource=social.tw" [X Link](https://x.com/leonpalafox/status/1967456020613214706) 2025-09-15T05:10Z [----] followers, [---] engagements "@scaling01 Thats between 83k and 200k way above what academia pays for similarly skilled people" [X Link](https://x.com/leonpalafox/status/1967554415541518835) 2025-09-15T11:41Z [----] followers, 26.3K engagements "@askewered @scaling01 Yeah but what I meant is that OP is claiming [--] usd/hr is peanuts for someone from IMO when thats the minimum rate for people in Academia" [X Link](https://x.com/leonpalafox/status/1967562124705890584) 2025-09-15T12:12Z [----] followers, [----] engagements "@Dan_Jeffries1 I hardly think that's Kimmel's case and is relevant because that is the trigger for all of this discussion he is leaving with a golden parachute to live in his mansion not going to a gulag to mine salt" [X Link](https://x.com/leonpalafox/status/1969044985648402936) 2025-09-19T14:24Z [----] followers, [---] engagements "@dhh is the box setting move your hands as if they were in a box Bill Clinton was one of the first to start doing it" [X Link](https://x.com/leonpalafox/status/1969051762498720042) 2025-09-19T14:51Z [----] followers, [---] engagements "Theres around 300k J1 visas granted each year and around and around 120k H1B out of which around 70% are granted to Indian consultancy companies. Saying that H1B will hamper researchers is untrue. Most academics in prestigious institutions come from other institutions within the US come from other institutions within the US which means they followed the F- OPT-O1" [X Link](https://x.com/leonpalafox/status/1969523379087495527) 2025-09-20T22:05Z [----] followers, [---] engagements "Depends. When we train models theres this thing called the loss function and thats usually what you tracks in the first round of pre training thats where a lot of the compute is going. Just adding more compute would mean you can do it faster so you can try different variations faster" [X Link](https://x.com/leonpalafox/status/1971957390711865537) 2025-09-27T15:17Z [----] followers, [--] engagements "@jasminewsun @natolambert When people says "Yann is no longer pro LLM" I'm like.that was never his main line of research he has been pushing for energy based model for decades now" [X Link](https://x.com/leonpalafox/status/1971967803285819852) 2025-09-27T15:58Z [----] followers, [--] engagements "@giffmana The whole pint of the comic is that for Prof Smith unless it worked it was a unproductive week which is the crux of many students and why they spend uncountable hours in the lab bad professors expect good results all the time" [X Link](https://x.com/leonpalafox/status/1973008517205459037) 2025-09-30T12:54Z [----] followers, [---] engagements "@Noahpinion Youre making a terrible mistake. This is Buenos Aires" [X Link](https://x.com/leonpalafox/status/1975608327230165217) 2025-10-07T17:05Z [----] followers, [----] engagements "Si les gust KPop Demon Hunters permtanme introducirlos a la palabra Santa de Macross Frontier" [X Link](https://x.com/leonpalafox/status/1976513090083521012) 2025-10-10T05:00Z [----] followers, [---] engagements "@Mathinity_ Lol ChatGPT firs stab was with numerical methods" [X Link](https://x.com/leonpalafox/status/1979205171537141803) 2025-10-17T15:17Z [----] followers, [----] engagements "There are a bazillion papers on crater detection yet no one shares their code Space Science certainly has terrible practices" [X Link](https://x.com/leonpalafox/status/674283464640737281) 2015-12-08T17:44Z [----] followers, [--] engagements "Some business are just out there to take advantage of people. charges you [--] USD to check if you need a Visa. http://travisa.com http://travisa.com" [X Link](https://x.com/leonpalafox/status/771505615445168129) 2016-09-02T00:30Z [----] followers, [--] engagements Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing
@leonpalafox Leon PalafoxLeon Palafox posts on X about ai, hay, aos, cuando the most. They currently have [-----] followers and [---] posts still getting attention that total [------] engagements in the last [--] hours.
Social category influence technology brands 15.28% countries 6.94% stocks 5.56% currencies 4.17% finance 3.47% social networks 3.47% automotive brands 2.08% celebrities 1.39% cryptocurrencies 0.69% gaming 0.69%
Social topic influence ai 4.86%, hay 4.17%, aos 3.47%, cuando 3.47%, gente 3.47%, cmo 2.78%, vibe coding 2.78%, eso 2.78%, $googl #1484, anthropic 2.78%
Top accounts mentioned or mentioned by @osvaldoags @billymannino @danielmac8 @jordineil @zocojonn @jesusvzr @deanmckee757 @wordgrammer @catherder1685 @thetransitguy @grok @jessicobra @noseofdeath @pfau @danjeffries1 @scaling01 @luisjacob_gg @mxquiquebm @sammoga123 @claro_vi33804
Top assets mentioned Alphabet Inc Class A (GOOGL) Microsoft Corp. (MSFT) GrokCoin (GROKCOIN) Tesla, Inc. (TSLA) Solana (SOL)
Top posts by engagements in the last [--] hours
"@jack_ceroni Learn to use AI do not shy away from it"
X Link 2026-02-14T18:27Z [----] followers, [---] engagements
"Ok no se si es meme pero esto parece mas apocaliptico que nada mas JUST IN: NYC bar opens for people with "AI" partners dining tables optimized to dine with your AI companion. JUST IN: NYC bar opens for people with "AI" partners dining tables optimized to dine with your AI companion"
X Link 2026-02-14T18:28Z [----] followers, [---] engagements
"Cmo empezar con vibe coding en ciencia de datos Paso [--] Empieza con un proyecto que domines. Conoces el modelo sus sesgos cmo se optimiza y entiendes bien los datos. Para qu Para poder validar lo que te devuelve el agente. Si algo no cuadra o sale mal lo detectas de inmediato. Paso [--] Ahora s deja que el agente modifique o agregue cosas pero siempre dentro del alcance de tu entendimiento. Aqu aprovechas que escribe cdigo mucho ms rpido que t. Eso s: pdele siempre grficas mtricas y seales que te permitan validar lo que est haciendo. Paso [--] Con un baseline slido deja que el agente se ponga"
X Link 2026-02-03T15:43Z [----] followers, [----] engagements
"Y Jsto regresa a operar con nuevos rostros y con nuevo dinero. Igual que lo coment hace algunos aos el modelo de negocio necesita cambiar si en realidad quieren seguir vivos. En retail la divisin que histricamente hace menos dinero es la de perecederos (en Walmart nadie espera hacer dinero con perecederos) y es justo el nicho de Justo. Encima de eso tienen la ltima milla de perecederos histricamente la logstica ms cara. Veamos que sorpresa nos traen en mi opinin necesitan modelos muy precisos de forecast e investigacin de operaciones para tratar de ser rentables pero ya veremos"
X Link 2026-02-04T12:14Z [----] followers, 13.1K engagements
"22 aos despus nadie nunca me ha enseado una visualizacin de R que no se pueda hacer con Python. Y ahora con VibeCoding menos"
X Link 2026-02-04T21:54Z [----] followers, 34.4K engagements
"Si viste los resultados de Microsoft y Google es fcil quedarse con el crecimiento (Google Cloud al 48% Azure al 39%). Impresionante s. Pero no es lo ms importante. La historia real es la chequera. Estamos viendo cmo se construye el foso econmico ms caro que hemos visto. El CapEx ya no es alto es brutal: Microsoft meti USD [----] mil millones en un solo trimestre y Alphabet est hablando de USD [------] mil millones para [----]. Eso es ms de lo que produce un pas entero en un ao. La guerra del cloud ya no va de software ni de ventas. Va de quin puede controlar el silicio y la capacidad de cmputo. En"
X Link 2026-02-05T01:56Z [----] followers, [----] engagements
"Eres junior data scientist y sientes que el vibe coding te est quitando el trabajo No ests solo. Es una realidad: muchas empresas estn contratando menos juniors y en cambio estn dando mejores herramientas a desarrolladores ms senior. Entonces qu toca hacer Volverte ms indispensable que el cdigo que generas. No es solo hacer una query o correr un anlisis es entender lo que ests construyendo y por qu importa. Si tu valor es escribir una query y entregar un dashboard s: eres fcil de reemplazar. Pero si tu valor es hablar con el negocio coordinar equipos mover engranes y llevar conversaciones con"
X Link 2026-02-05T12:19Z [----] followers, 20.7K engagements
"Con todo el tema del vibe coding creciendo usar lenguaje natural para construir software Anthropic lanza Claude Opus [---] y se nota el salto. Mejor manejo de contexto largo ms estabilidad en tareas complejas y una experiencia mucho ms fluida para acompaar flujos reales de trabajo. Se siente ms slido cuando lo usas varias horas seguidas y no solo para generar snippets. Cada vez queda ms claro dnde est el valor: tener claridad sobre qu ests construyendo cmo se conecta con el sistema y cmo validar resultados. El reto pasa por el criterio y la comprensin del problema."
X Link 2026-02-06T00:25Z [----] followers, [----] engagements
"Alguien por ac me peleaba que no se podra construir un Kernel desde cero con AI Anthropic had [--] AI agents build a C compiler from scratch. 100k lines compiles the Linux kernel $20k [--] weeks. To put that in perspective GCC took thousands of engineers over [--] years to build. (Granted from [----] - however) One researcher and [--] AI agents just built a https://t.co/pk0PxsdTKi Anthropic had [--] AI agents build a C compiler from scratch. 100k lines compiles the Linux kernel $20k [--] weeks. To put that in perspective GCC took thousands of engineers over [--] years to build. (Granted from [----] - however)"
X Link 2026-02-06T02:18Z [----] followers, 20.8K engagements
"Y el que faltaba present ya tenemos resultados de Amazon Web Services (AWS) Google y Microsoft . Es sin lugar a duda una de las inversiones privadas ms importantes desde la poca de los ferrocarriles. Y como siempre la nube es un bien pblico"
X Link 2026-02-06T22:34Z [----] followers, [---] engagements
"Yo tuve y conoc varios profesores que se vanagloriaban de reprobar alumnos o de lo difcil que era pasar su clase. Aos despus sigo convencido de lo mismo: ese tipo de personas no deberan ser maestros. Deberan dedicarse a otra cosa menos a formar a nuevas generaciones. Porque su objetivo nunca fue educar sino generar frustracin. Tu trabajo no es reprobar gente con exmenes para los que no los preparaste. Tu trabajo es asegurarte de que la mayora de tus alumnos entiende el material que impartes. Presumir que pocos pasan tu clase no es seal de excelencia acadmica. Es casi siempre una forma"
X Link 2026-02-09T12:14Z [----] followers, 23.7K engagements
"Esta grfica lo nico que muestra es que se est volviendo a la tranquilidad despus de que durante COVID te daban un puesto con que supieras hacer un import"
X Link 2026-02-09T18:50Z [----] followers, [----] engagements
"@data_complex No es verdad el esfuerzo de un modelo mayoritario en espaol no es trivial y no se resuelve con un fine tunning"
X Link 2026-02-09T22:20Z [----] followers, [---] engagements
"@daniel_mac8 Dude be careful Gary threatens litigation at the smallest slight"
X Link 2026-02-10T18:25Z [----] followers, [---] engagements
"Esto estimados es lo que en verdad puede adelantar a la humanidad varios aos Pharma is COOKED Isomorphic Labs just revealed IsoDDE: an AI system that designs drugs on a computer faster than any pharma R&D doubles AlphaFold [--] on hard targets 20x better than Boltz-2 on antibodies beats the physics gold standard at binding found drug pockets from https://t.co/iQ3xKqU9mw Pharma is COOKED Isomorphic Labs just revealed IsoDDE: an AI system that designs drugs on a computer faster than any pharma R&D doubles AlphaFold [--] on hard targets 20x better than Boltz-2 on antibodies beats the physics gold"
X Link 2026-02-10T18:40Z [----] followers, [----] engagements
"Existe mucha gente que la IA es una especie de base de datos donde en el momento que le preguntas algo va y trae el resultado. La realidad es muy distinta si funcionara as le llamaramos base de datos qu padre llamarle "pensar" a encontrar la misma tesis de la que hablas en una base de datos scrappeada sin permiso de nadie y regurgitarte el resultado que a ti te cost meses de trabajo qu padre llamarle "pensar" a encontrar la misma tesis de la que hablas en una base de datos scrappeada sin permiso de nadie y regurgitarte el resultado que a ti te cost meses de trabajo"
X Link 2026-02-11T13:52Z [----] followers, [----] engagements
"Pens que ya habamos superado esta discusin pero parece que vale la pena insistir: La IA no es una base de datos. No funciona como un repositorio al que le metes un libro y luego simplemente lo escupe completo cuando alguien lo pide. De hecho es extremadamente difcil que reproduzca citas largas de forma exacta incluso si el texto form parte de su entrenamiento. Lo que hace es distinto: aprende patrones estadsticos del lenguaje y de las estructuras matemticas. No memoriza respuestas como si fueran archivos; construye salidas a partir de representaciones internas. Y esto no es teora. Ya existen"
X Link 2026-02-11T14:38Z [----] followers, 222.6K engagements
"La gente ignora vehementemente que el problema no fue que sus inversiones fueran buenas o malas sino que lo estaba haciendo con dinero de otras personas que no le aprobaron hacer eso. Next time you feel bad for missing something that mooned Imagine being SBF: - Put $500m into Anthropic which would now be worth $70b. - Held $60m in $SOL at $8 that would've been worth $2.1b at the top. - Put $100m into Mysten Labs that would now be worth over $800m. - https://t.co/6axraF18DC Next time you feel bad for missing something that mooned Imagine being SBF: - Put $500m into Anthropic which would now be"
X Link 2026-02-11T14:49Z [----] followers, [----] engagements
"@JordiNeil Es bien sabido que hay una paradoja en Deep Learning donde es muy bueno generalizando aun cuando se le entren para hacer overfitting https://lis.csail.mit.edu/pubs/kawaguchi-techreport18.pdf https://lis.csail.mit.edu/pubs/kawaguchi-techreport18.pdf"
X Link 2026-02-11T15:02Z [----] followers, [----] engagements
"Las respuestas a este post me indican que si que debemos segur discutiendo esto. La poblacin realmente cree que sabe de tecnologa cuando no tiene idea de la diferencia entre ; - Motor de Bsqueda - Base de Datos - LLM Creen que se pueden usar como palabras intercambiables y no lo son Pens que ya habamos superado esta discusin pero parece que vale la pena insistir: La IA no es una base de datos. No funciona como un repositorio al que le metes un libro y luego simplemente lo escupe completo cuando alguien lo pide. De hecho es extremadamente difcil Pens que ya habamos superado esta discusin pero"
X Link 2026-02-12T13:28Z [----] followers, [----] engagements
"Con los avances que he visto no me sorprendera que escribir cdigo muera. Es una tarea sper mecnica donde tenemos que poner capas de abstraccin para poder hacer algo til (si un lenguaje de programacin es una abstraccin para humanos) Con suficiente informacin en binario y hexadecimal podemos crear IAs que puedan crear cualquier cosa escrita directamente en binario y toda interaccin humana terminara. Lo que dudo mucho es que suceda en los prximos [--] aos y menos en este ao Elon Musk thinks coding dies this year. Not evolves. Dies. By December AI wont need programming languages. It generates"
X Link 2026-02-12T14:40Z [----] followers, 13.5K engagements
"@claro_vi33804 Fabuloso tu poder de discusin"
X Link 2026-02-12T20:14Z [----] followers, [--] engagements
"Machine Learning sucks. La frase es de Yann LeCun. Y aunque suena exagerada tiene fondo. Nuestros modelos actuales son impresionantes pero profundamente ineficientes. Aprenden con ocanos de datos lo que la inteligencia biolgica capta con unas cuantas experiencias. En Duke LeCun insisti en algo incmodo: ms datos no es igual a AGI. Los LLMs autoregresivos predicen token por token; mientras ms larga la respuesta mayor la probabilidad de desviarse. Escalan s. Pero no necesariamente razonan. Por eso su empuje hacia JEPA y world models no es capricho. Es un cambio de direccin: menos obsesin por"
X Link 2026-02-12T21:56Z [----] followers, [----] engagements
"Yo he estado a punto de pagar los [---] dlares al mes ms de una vez ever have certain moments where the only thing in your mind is: "what the fuck" i mean you gotta admire the narrative discipline the execution & the go to market. this is incredible. https://t.co/Adh8jpXiBq ever have certain moments where the only thing in your mind is: "what the fuck" i mean you gotta admire the narrative discipline the execution & the go to market. this is incredible. https://t.co/Adh8jpXiBq"
X Link 2026-02-13T00:42Z [----] followers, [----] engagements
"En nuestro episodio de ayer de la clase de Deep Learning: Hay una pregunta muy sencilla que suele revelar si alguien realmente entiende cmo funcionan los modelos de lenguaje: Puede explicar la diferencia entre un token y un embedding Un token es la unidad bsica con la que el modelo procesa texto. Es cmo fragmenta lo que le damos y lo que genera. Ah se miden costos lmites de contexto y velocidad. Un embedding es la representacin numrica del significado. Es el momento en que el lenguaje se convierte en un vector dentro de un espacio matemtico donde las relaciones semnticas se pueden medir."
X Link 2026-02-13T17:27Z [----] followers, 14.6K engagements
"@zocojonn No hay mucho ingeniero de IA que sabe eso pero no sabe exactamente cmo se calcula un embedding o qu cosa es gradiente descendiente esos trminos son ms de aplicaciones"
X Link 2026-02-14T15:26Z [----] followers, [--] engagements
"La gente cree que esto con ciclovas se arregla pero no lo principal es transporte pblico. Esto es un problema que le debemos a administraciones de izquierda que en [--] aos nos han dado una sola lnea de metro y hasta el metrobs (que es un parche) parece que se les olvid en pos de trolebuses. Brugada sigue presumiendo que va a haber dos lneas nuevas de la cual se ve cero avance La CDMX sigue siendo de las ciudades ms contaminadas del continente y con mayor trfico EN EL MUNDO y esto no es casualidad. Y todo a causa de sus polticas de abandonar la expansin del metro sustituirlo por autobuses y"
X Link 2026-02-14T16:46Z [----] followers, [----] engagements
"@OsvaldoAGS Te invito a ver cuantas ciclovas hay en Tokio u Osaka"
X Link 2026-02-14T17:04Z [----] followers, [---] engagements
"@moaimx Que modelo estas usando La mayora de los foros dicen que Claude no tiene mayor tema"
X Link 2026-02-14T18:30Z [----] followers, [--] engagements
"Existe una clara estrategia para hacer frente a la IA Estamos entrando a una nueva divisin social frente a la inteligencia artificial. No es solo quin la usa y quin no. La diferencia real est en la combinacin entre mentalidad y capacidad. Hoy conviven cuatro perfiles: Quienes quieren aprender pero no tienen acceso a herramientas. Quienes tienen acceso pero la desestiman. Quienes ni tienen acceso ni inters. Quienes quieren aprender y adems se equipan para usarla estratgicamente. La evidencia sobre adopcin tecnolgica muestra que la ventaja no proviene solo del acceso ni solo de la intencin."
X Link 2026-02-14T18:46Z [----] followers, [----] engagements
"@OsvaldoAGS @JesusVZR Me podras por favor compartir el link donde muestranq ue mueven ms gente para el caso de la CDMX (y no una ciudad Europea con muchos menos habitantes)"
X Link 2026-02-14T18:47Z [----] followers, [--] engagements
"Anthropic hitting a $14B run rate in just three years is incredible to see. The real signal here is Claude Code: its already at a $2.5B run rate and driving 4% of all GitHub commits. We are watching the developer landscape shift in real-time"
X Link 2026-02-13T00:48Z [----] followers, [---] engagements
"Esto es interesante y muy real. Asi como hoy hay personas que "pues hago lo que dice en la maquina" y otras que "hacen lo que ponen en la maquina" In [--] years there will be two classes of people. Economists call it the "K-shaped economy" - and the next 2-3 years will decide which line you're on. An overclass that uses AI as a lever to build wealth automate income and make decisions at a speed no human can compete https://t.co/pN8QZBz9cv In [--] years there will be two classes of people. Economists call it the "K-shaped economy" - and the next 2-3 years will decide which line you're on. An"
X Link 2026-02-14T18:36Z [----] followers, [----] engagements
"La gran realidad es que el 80% de las cosas que se hagan con IA sern triviales The punchline here is incredible https://t.co/qoznql8t4X The punchline here is incredible https://t.co/qoznql8t4X"
X Link 2026-02-14T18:56Z [----] followers, [----] engagements
"Madre. [----] varos. podra probarlo un mes u ver que onda. The upgraded Deep Think mode is rolling out now in the @GeminiApp for Google AI Ultra subscribers. For scientific researchers and developers were opening a Vertex AI Early Access Program for the API. Start discovering https://t.co/8OhnV65pW5 https://t.co/WGSlntIFLc The upgraded Deep Think mode is rolling out now in the @GeminiApp for Google AI Ultra subscribers. For scientific researchers and developers were opening a Vertex AI Early Access Program for the API. Start discovering https://t.co/8OhnV65pW5 https://t.co/WGSlntIFLc"
X Link 2026-02-14T22:20Z [----] followers, 18.7K engagements
"Me causa curiosidad saber que fue lo que necesit el Pentgono particularmente de la IA de @AnthropicAI . Uno pensara que tienen suficiente informacin"
X Link 2026-02-15T06:03Z [----] followers, [----] engagements
"@Prazkat Dispararle a los gays en Los Feliz es una clara muestra de lo Woke que es"
X Link 2023-12-05T18:40Z [----] followers, 59.7K engagements
"La IA puede apoyar a descubrir nuevas cosas en la fisica. Ciertamente te puede ayudar a hacer la tarea GPT-5.2 derived a new result in theoretical physics. Were releasing the result in a preprint with researchers from @the_IAS @VanderbiltU @Cambridge_Uni and @Harvard. It shows that a gluon interaction many physicists expected would not occur can arise under specific GPT-5.2 derived a new result in theoretical physics. Were releasing the result in a preprint with researchers from @the_IAS @VanderbiltU @Cambridge_Uni and @Harvard. It shows that a gluon interaction many physicists expected would"
X Link 2026-02-14T01:47Z [----] followers, [----] engagements
"@OsvaldoAGS Mas del doble jajajaja"
X Link 2026-02-14T18:03Z [----] followers, [--] engagements
"@jmtoralc Lo ms pro es hacerlo en wortrees. Yo lo hice una vez y se acab me dinero en menos de una hr"
X Link 2026-02-14T22:09Z [----] followers, [---] engagements
"@fernetcontonica @futboltrotters En Mxico la mam de Oliver (nada de Tsubasa) le agradece a la virgen de Guadalupe"
X Link 2026-02-14T22:12Z [----] followers, [---] engagements
"Para ser honestos me gustara poder ir en tren al trabajo tambin Solo puedo decir que amo vivir en Toluca por el simple hecho de poder utilizar un tren para ir a la CDMX https://t.co/dFCxNI0WNC Solo puedo decir que amo vivir en Toluca por el simple hecho de poder utilizar un tren para ir a la CDMX https://t.co/dFCxNI0WNC"
X Link 2026-02-10T18:17Z [----] followers, 10.5K engagements
"Para m la mejor prueba del cido que alguien sabe o no de IA es que en verdad sabe la diferencia entre un Token y un Embedding"
X Link 2026-02-13T15:25Z [----] followers, 42.9K engagements
"Yo sigo mucho a Ray por LinkedIn y he ledo sus libros. Me parece una buena lectura aunque exagera en algunas cosas https://t.co/tjmbT5ytUN https://t.co/tjmbT5ytUN"
X Link 2026-02-15T23:59Z [----] followers, [----] engagements
"You can generate multiple parallel answers and start auto pruning the tree which is probably what they did for o3. Yann is talking about a single autoregressive run and he is right on that one the more tokens you generate the more likely it is to pick a low probability token and start going astray"
X Link 2024-12-23T14:21Z [----] followers, [---] engagements
"@deanmckee757 @wordgrammer What If you work in a bank you get tabular data as daily transactions. If you work at retail sales and client behavior also comes as tabular data"
X Link 2024-12-24T19:27Z [----] followers, [--] engagements
"People saying that GenAI can replace traditional ML are my favorite people right away you know theyre full of bull"
X Link 2024-12-24T19:29Z [----] followers, [---] engagements
"@deanmckee757 @wordgrammer And why would you spend that much compute to solve marginally better something that XGBoost can solve Doesnt make an iota of sense I used to score millions of clients each week. Saying that is better to use GenAI instead does not make much sense"
X Link 2024-12-24T19:53Z [----] followers, [--] engagements
"@Lons Just tell them is the sequel of Brad Pitt movie"
X Link 2024-12-25T03:28Z [----] followers, [--] engagements
"Im sure those kids will remember their folks grinding to make Grok tell slightly less wrong answers and not the good time they could have spent with their parents. At @xai we were planning to grind through Christmas. But then @ibab had a strange experience with three apparitions 👻👻👻. The regular grinding will continue after tomorrow. At @xai we were planning to grind through Christmas. But then @ibab had a strange experience with three apparitions 👻👻👻. The regular grinding will continue after tomorrow"
X Link 2024-12-25T07:22Z [----] followers, [---] engagements
"@bryan_johnson I mean 90% of people still dont understand the Internet"
X Link 2024-12-25T16:12Z [----] followers, [--] engagements
"@billymannino Even at that Apple is also digital which means there is a quantization involved which means theres a loss. Unless is bulbs all the way from recording to output there are losses"
X Link 2024-12-26T22:19Z [----] followers, 22K engagements
"@Catherder1685 @billymannino Yes but thats not the point the point is that there isnt a truly 100% fidelity recording"
X Link 2024-12-27T05:30Z [----] followers, [---] engagements
"@OCopium @billymannino Speaker has a piezoelectric piece that transforms current into vibrations the speaker in and of itself offers way more fidelity than the AC/DC Also being driven by electronics doesnt imply loss of fidelity as you can have analog electronics. And not all electronics are digital"
X Link 2024-12-27T14:50Z [----] followers, [---] engagements
"@Catherder1685 @billymannino Not really since my first post Ive been saying that quantization in a digital device implies that there is no 100% fidelity"
X Link 2024-12-27T14:51Z [----] followers, [--] engagements
"@StevenPulteFam Why spend 4k in your 20s for a watch Put it in the SP500 and buy a car in your 40s"
X Link 2024-12-29T01:58Z [----] followers, [---] engagements
"@the_transit_guy lol comparing Spain (a whole country) with Houston a single city"
X Link 2024-12-30T01:03Z [----] followers, [---] engagements
"@nkalamb @nickjanetos @Rider_Nytimus @_Zeets I have a hard time thinking that a ton of very educated scholars got the translations wrong but she got it right"
X Link 2024-12-30T14:34Z [----] followers, [----] engagements
"@heartsalty I dont think Fagles or Lattimore were just bouncing from other translations"
X Link 2024-12-30T15:02Z [----] followers, [---] engagements
"Image by @grok"
X Link 2024-12-30T23:33Z [----] followers, [--] engagements
"@HanchungLee In every job I've had we had a team of DSs and a team of DEs so one would experiment in Jupyter/Rstudio and the opther would create the pipelines is Scala/Spark/Databricks"
X Link 2024-12-31T15:00Z [----] followers, [---] engagements
"@SyncereX8 @kingofbrethren What's wron with Brad Pitt as a Mexican"
X Link 2024-12-31T15:53Z [----] followers, [----] engagements
"@lemire Go further if you had invested [--] USD in Intel in the [----] you would have exactly the same amount of money and if you had done it in [----] you've lost [--] times your money"
X Link 2024-12-31T21:35Z [----] followers, [---] engagements
"The people wondering why Grok [--] hasn't shipped.did you know what happened when Musk promised Teslas for a certain day Musk is terrible at reaching deadlines because he overpromises"
X Link 2025-01-03T15:18Z [----] followers, [---] engagements
"@sergiooliveiram Nissan las Peugeot se les funde un faro y se les prenden mil testigos"
X Link 2025-01-04T15:58Z [----] followers, [--] engagements
"@alexrmoskowitz The only function of economic forecasting is to make astrology look respectable"
X Link 2025-01-05T23:51Z [----] followers, [----] engagements
"@GoodPoliticGuy Noup it was the overall willingness of the Dems to not look the rural bit and focus instead on the college educated"
X Link 2025-01-06T00:38Z [----] followers, [----] engagements
"@Coolio_Art Those arent accurate raptors they are way too big"
X Link 2025-01-06T05:04Z [----] followers, [---] engagements
"@Beach_Wolf69 @WomanDefiner I think he makes a better service exposing these kind of spenders. To show that even people who win [--] figures (in this case a couple) can blow all their money and have a hard time getting out of debt"
X Link 2025-01-07T14:57Z [----] followers, [----] engagements
"@DarkwebShaman In Japan they don't have black people and they still have those doors"
X Link 2025-01-08T14:22Z [----] followers, [--] engagements
"@jessicobra How is Capitalism bad"
X Link 2025-01-08T14:27Z [----] followers, [----] engagements
"@hercine3 @jessicobra No I'm saying that the acumulaton of wealth is not a feature of capitalism is a feature of human nature. Thus your one complain about capitalism is not exclusive of it"
X Link 2025-01-08T16:14Z [----] followers, [---] engagements
"@ajordannafa You cant use SGd if the matrix X is millions of element large computationally speaking is way more effective to use SGD"
X Link 2025-01-09T04:53Z [----] followers, [---] engagements
"@nhfoley @bryancsk All of Mexico City is concrete buildings in a VERY earthquake prone zone"
X Link 2025-01-12T05:27Z [----] followers, [--] engagements
"I love AI Data Teams where there isn't a single person that really knows about AI"
X Link 2025-01-13T19:05Z [----] followers, [---] engagements
"@GayBearRes In many companies this is the case. Im seeing it as well. Most of our consultants (Accenture Indigene etc) have their tech teams based in India"
X Link 2025-01-14T17:06Z [----] followers, [--] engagements
"And maybe she is being paid by TikTok to say this. This video by popular TikTok personality Soupy is going viral with [---] million likes and [---] million views in [--] hours. Gen z is not happy. https://t.co/bhfCPv1do7 This video by popular TikTok personality Soupy is going viral with [---] million likes and [---] million views in [--] hours. Gen z is not happy. https://t.co/bhfCPv1do7"
X Link 2025-01-14T17:11Z [----] followers, [---] engagements
"Theres a big difference Bad Bunny wants to cater to an American audience but youre not seeing Fallon in Telemundo Bad Bunny telling Jimmy Fallon how he's improved his English over the years and that it's Fallon's turn to improve his Spanish is EXACTLY what these gringos need to hear. They always expect to be catered to and get uncomfortable when they aren't. Bad Bunny telling Jimmy Fallon how he's improved his English over the years and that it's Fallon's turn to improve his Spanish is EXACTLY what these gringos need to hear. They always expect to be catered to and get uncomfortable when they"
X Link 2025-01-14T23:11Z [----] followers, [---] engagements
"@kopeboy @pitdesi The Disney parks revenue is about [--] billion USD a year so .not much"
X Link 2025-01-16T11:49Z [----] followers, [--] engagements
"@LeeWinterOz @pot8um Because it was hard from the developer POV few people remember the hell that was the yearly release cycle now you have incremental upgrades"
X Link 2025-01-16T13:23Z [----] followers, [---] engagements
"@ZebitasMartinex Convenientemente te saltaste VirtualBoy N64 y Game Cube"
X Link 2025-01-17T04:31Z [----] followers, [---] engagements
"@445543636536s @the_transit_guy Also they are comparing a 450million continent with a [---] million country Just by virtue of that is obvious main cities will have more people Go and compare with India or China"
X Link 2025-01-18T13:49Z [----] followers, [--] engagements
"Since the Wii every time Nintendo releases a new console haters talk about graphics. People dont realize that Nintendo already went the path of latest graphics chipset and it failed miserably. The N64 was more powerful than the PS One and the Cube was more powerful than the PS2"
X Link 2025-01-20T03:41Z [----] followers, [----] engagements
"@Logan96194157 @NoseOfDeath The Cube used CDs"
X Link 2025-01-20T11:33Z [----] followers, [---] engagements
"Sama is pretty much Musk [---] he has been hyping GenAI for the past month and now he is like uhhh no no that is justo Twitter being Twitter twitter hype is out of control again. we are not gonna deploy AGI next month nor have we built it. we have some very cool stuff for you but pls chill and cut your expectations 100x twitter hype is out of control again. we are not gonna deploy AGI next month nor have we built it. we have some very cool stuff for you but pls chill and cut your expectations 100x"
X Link 2025-01-20T12:40Z [----] followers, [---] engagements
"@GdGoncalo @NoseOfDeath Yeah no one argues that the storage medium in PS1 allowed for CG videos which the N64 sorely lacked but game rendering was superior in the N64"
X Link 2025-01-20T18:54Z [----] followers, [--] engagements
"@B2RKN @spectatorindex All Asian countries have very aggressive laws against drugs. Singapore one of the most advanced cities has the death penalty for possession of many drugs"
X Link 2025-01-21T02:22Z [----] followers, [---] engagements
"Tan fcil que era para Musk no hacer el saludo y ya Found this in Reddit seems like it might be useful for some people. https://t.co/Mg6tCg6aAB Found this in Reddit seems like it might be useful for some people. https://t.co/Mg6tCg6aAB"
X Link 2025-01-23T17:55Z [----] followers, [---] engagements
"@__Siffa @filippie509 3% Llama is estimated to have been trained on 16K processors also a ton of the hardware goes towards inference not just training"
X Link 2025-01-27T21:33Z [----] followers, [--] engagements
"@quesadillino @YosoyPedrero Fridays quebr"
X Link 2025-02-02T16:04Z [----] followers, [---] engagements
"Yet another GPT wrapper that companies want to pass as an innovative feature whyshould I pay you [--] USD for something I can do in my GPT account"
X Link 2025-02-04T16:48Z [----] followers, [---] engagements
"Esto pasa para todo no solo con atletas. Dios est ah para nuestras victorias pero cuando pasa una desgracia jams lo culpamos Curious that talented athletes frequently credit God when they win but we rarely see them blame God when they lose. Curious that talented athletes frequently credit God when they win but we rarely see them blame God when they lose"
X Link 2025-02-10T20:44Z [----] followers, [---] engagements
"This week in class we explored how LLMs are becoming the foundation for enterprise AI apps and what tools and concepts are powering that shift. Heres the architecture we discussed: 🔗 Taxonomies help structure knowledge and bring consistency to LLM interactions essential for governance 📚 RAG (Retrieval-Augmented Generation) connects LLMs to external knowledge sources to improve accuracy and reduce hallucination 🤖 Agents allow models to take actions use tools and handle multi-step tasks 🧩 Model Context Protocol (MCP) a new standard from Anthropic that lets multiple applications and tools"
X Link 2025-03-21T13:39Z [----] followers, [---] engagements
"@ostrisai Because it relates to the user and avoids paying a fee to do it or adding [--] thousand selfies. Same reason things are popular"
X Link 2025-03-28T14:54Z [----] followers, [--] engagements
"@paulg @mattyglesias Any random Chinese city looks better than NYC services public transport everything"
X Link 2025-03-31T15:32Z [----] followers, [----] engagements
"@skirano RAG was never about context was about reducing hallucinations. Long context does nothing to do that"
X Link 2025-04-05T22:43Z [----] followers, [----] engagements
"Madagascar.twice Delete one franchise FOREVER https://t.co/ojBZeIUOfE Delete one franchise FOREVER https://t.co/ojBZeIUOfE"
X Link 2025-04-10T20:18Z [----] followers, [---] engagements
"@My1xT @atom_jarvis @BridgetCambria8 @FrankLuntz Most CRM do the same thing they warm up the mails because if GMAIL detects a massive splurge of mails from the same address they will send them to junk mail"
X Link 2025-04-13T01:49Z [----] followers, [---] engagements
"@JesseSchoberg @levelsio This is disingenuous AB InBeV operates multiple brands in multiple countries its an amalgamation of many companies"
X Link 2025-04-13T17:08Z [----] followers, [----] engagements
""This order promotes integrating AI into K-12 education by encouraging AI literacy teacher training and partnerships with private and academic institutions. It establishes a White House Task Force on AI Education and directs federal agencies to prioritize AI in discretionary grant programs for teacher training""
X Link 2025-04-27T01:34Z [----] followers, [--] engagements
"@jxmnop The fix is easy do as any other science and accept pretty much every paper and make conferences a networking opportunity"
X Link 2025-05-04T04:24Z [----] followers, [---] engagements
"@ICBItsNotBetty @ArmandDoma No I decided to go to Asia to study robotics in a plan thats way better than anything Harvard has"
X Link 2025-05-26T02:25Z [----] followers, [--] engagements
"@hecate40 @kareem_carr CEO of Meta - Grandson of immigrants CEO of Tesla - Immigrant CEO of Google - Immigrant CEO of OpenAI - also from a family of immigrants. Man I wonder how can America innovation can get down if you stop preventing immigrants from going to your country"
X Link 2025-05-28T11:46Z [----] followers, [---] engagements
"These three legends are basically the north star I follow when it comes to story telling humor and energy"
X Link 2025-06-09T16:40Z [----] followers, [---] engagements
"Were quick to criticize ChatGPT for killing our critical thinking. But where was this outrage when we spent years normalizing rote memorization as the default way of teaching Many of us were forced to memorizeby heartthe European countries and their capitals. How much of that knowledge do we really retain today And how many schools still teach math the same way history is taught: by reciting formulas without understanding how or why they work Lets be honest. Long before ChatGPT we already had a school system that often discouraged curiosity punished questioning and prioritized repetition over"
X Link 2025-06-19T05:01Z [----] followers, [---] engagements
"@MushtaqBilalPhD Academic Publishers is one of the greates transferrs of cash from Government to private coorporations that we've ever seen. If anyone saw ti from that lens I'm sure governments would stop funding these"
X Link 2025-06-22T17:44Z [----] followers, 12K engagements
"@ankurnagpal Zoom tried to compete against Microsoft in their own game and of course no one can challenge MSFT distributionc apabilities. It literally had zero edge against Teams"
X Link 2025-06-25T00:16Z [----] followers, [---] engagements
"@jboysen0 @pfau I refer to CS only bros who know shit about wet experiments yet want to change health and medicine"
X Link 2025-07-01T05:02Z [----] followers, [---] engagements
"@haro_ca_ Why"
X Link 2025-07-08T18:57Z [----] followers, [--] engagements
"@pfau Whats even the point Is like having AI art created to be attended solely by robots"
X Link 2025-07-11T15:52Z [----] followers, [--] engagements
"He has a PhD in Machine Learning and needed an expert to tell him using Transformers for tabular data is a bad idea"
X Link 2025-07-23T14:08Z [----] followers, [---] engagements
"Someone should make the experiment try fine tuning on all last reviews and then have a cohorts of humans reviewing and an LLM reviewer and see how much acceptance rates change NeurIPS reviews are now complete trash. NeurIPS reviews are now complete trash"
X Link 2025-07-25T14:19Z [----] followers, [---] engagements
"@2prime_PKU Hey at least you know he or she didnt use an LLM 😆"
X Link 2025-07-25T17:20Z [----] followers, [---] engagements
"Hay dos chief scientists de AI en meta We are excited to announce that @shengjia_zhao will be the Chief Scientist of Meta Superintelligence Labs Shengjia is a brilliant scientist who most recently pioneered a new scaling paradigm in his research. He will lead our scientific direction for our team. Let's go 🚀 https://t.co/D93KQWIvFl We are excited to announce that @shengjia_zhao will be the Chief Scientist of Meta Superintelligence Labs Shengjia is a brilliant scientist who most recently pioneered a new scaling paradigm in his research. He will lead our scientific direction for our team."
X Link 2025-07-26T00:14Z [----] followers, [---] engagements
"Dime que no sabes nada de la industria de autos sin decirme que no sabes nada de la industria"
X Link 2025-07-27T14:45Z [----] followers, [--] engagements
"Who wants to arrive [--] before to the stations and in [--] you are already in the train. Nooo I need to arrive [--] hours early either lo LAX or SFO (Both are a mess) lineup with strangers and take my shoes off and pc from my bag eat expensive airport food and theeeen get in my plane"
X Link 2025-07-31T14:19Z [----] followers, 12.2K engagements
"@MyPronounsRFOff @NiMingDa_888 Please tell me who did Venezuela sold most of its oil to"
X Link 2025-08-05T05:00Z [----] followers, [---] engagements
"@LinkofSunshine Average return rates are 10% thats above inflation and that gives you to live comfortably with 80k and leave the money growing until youre [--] to move it to a lower risk fund"
X Link 2025-08-05T17:38Z [----] followers, [----] engagements
"@JUSTDEMITOY Shakira stopped being a writer a long time ago most of her hits are pop/rap collaborations a long shot from she first started. She does write from time to time but not like other not established ones"
X Link 2025-08-08T17:20Z [----] followers, [---] engagements
"Is this the moment OpenAI jumps the shark"
X Link 2025-08-08T20:02Z [----] followers, [---] engagements
"95% of GenAI pilots are failing. Why All the problems weve seen for years in Data Science dirty data unguided models lack of clear business cases and no usage metrics didnt go away with Generative AI. In fact GenAI only amplified these issues. If your traditional ML models couldnt survive in production theres no way a GenAI model will. Most failures are not about the models themselves but about deeper structural problems in organizations. My advice Start with clear measurable business use cases. Define and standardize your prompts and architectures. Track usage and adoption from day one."
X Link 2025-08-21T19:56Z [----] followers, [---] engagements
"@dinels12 Ya varios te lo dijeron la IA que usa OpenAI es DeepLearning que es una rama de Machine Learning que en realidad no es ms que un rebranding de Redes Neuronales. Agarra un libro y ve eso"
X Link 2025-08-21T20:51Z [----] followers, [---] engagements
"Link: https://cloud.google.com/blog/products/infrastructure/measuring-the-environmental-impact-of-ai-inference https://cloud.google.com/blog/products/infrastructure/measuring-the-environmental-impact-of-ai-inference"
X Link 2025-08-22T00:07Z [----] followers, [---] engagements
"@Dan_Jeffries1 These people have no idea how medicine or Law even works. There you have the godfather of AI Hinton saying that radiologists would be without jobs now. These are people that rarely have gotten out to a world outside of their computers"
X Link 2025-08-23T20:53Z [----] followers, [---] engagements
"@francoisfleuret Yes I always teach it like in my ML class. We really have snippets of the phenomenon we are trying to model and we have the hubris to think that our model will be able to map that space. Sometimes it works sometimes it doesnt"
X Link 2025-08-29T20:24Z [----] followers, [--] engagements
"In every fusion sensor system you need a failsafe most smart cars have radar or sonar for their autostop systems (Nissan BYD etc) because sonar gives you a very fail proof system of estimating distance to an incoming obstacle. Video does not. Theres also this video that shows how you can Willy Coyote a Tesla https://youtu.be/IQJL3htsDyQsi=rGOCPPgCrYEpTQgF https://youtu.be/IQJL3htsDyQsi=rGOCPPgCrYEpTQgF"
X Link 2025-08-31T05:35Z [----] followers, [---] engagements
"Quizas ek futuro del internet sea una interaccin masiva con LLMs que son conectados con fuentes fidedignas (periodicos gobiernos) i never took the dead internet theory that seriously but it seems like there are really a lot of LLM-run twitter accounts now i never took the dead internet theory that seriously but it seems like there are really a lot of LLM-run twitter accounts now"
X Link 2025-09-04T01:43Z [----] followers, [---] engagements
"@daniel_mac8 OpenAI just discovered what the rest of the community has known for years"
X Link 2025-09-06T17:20Z [----] followers, [---] engagements
"@Maav_BR @martinmbauer An LLM is a transformer. The real answer is that they did not use a transformer trained on internet text but on model results"
X Link 2025-09-11T23:43Z [----] followers, [---] engagements
"I just published OpenAIs $300 Billion Gamble on Unreliable Answers https://medium.com/p/openais-300-billion-gamble-on-unreliable-answers-3f3959da4bfdsource=social.tw https://medium.com/p/openais-300-billion-gamble-on-unreliable-answers-3f3959da4bfdsource=social.tw"
X Link 2025-09-15T05:10Z [----] followers, [---] engagements
"@scaling01 Thats between 83k and 200k way above what academia pays for similarly skilled people"
X Link 2025-09-15T11:41Z [----] followers, 26.3K engagements
"@askewered @scaling01 Yeah but what I meant is that OP is claiming [--] usd/hr is peanuts for someone from IMO when thats the minimum rate for people in Academia"
X Link 2025-09-15T12:12Z [----] followers, [----] engagements
"@Dan_Jeffries1 I hardly think that's Kimmel's case and is relevant because that is the trigger for all of this discussion he is leaving with a golden parachute to live in his mansion not going to a gulag to mine salt"
X Link 2025-09-19T14:24Z [----] followers, [---] engagements
"@dhh is the box setting move your hands as if they were in a box Bill Clinton was one of the first to start doing it"
X Link 2025-09-19T14:51Z [----] followers, [---] engagements
"Theres around 300k J1 visas granted each year and around and around 120k H1B out of which around 70% are granted to Indian consultancy companies. Saying that H1B will hamper researchers is untrue. Most academics in prestigious institutions come from other institutions within the US come from other institutions within the US which means they followed the F- OPT-O1"
X Link 2025-09-20T22:05Z [----] followers, [---] engagements
"Depends. When we train models theres this thing called the loss function and thats usually what you tracks in the first round of pre training thats where a lot of the compute is going. Just adding more compute would mean you can do it faster so you can try different variations faster"
X Link 2025-09-27T15:17Z [----] followers, [--] engagements
"@jasminewsun @natolambert When people says "Yann is no longer pro LLM" I'm like.that was never his main line of research he has been pushing for energy based model for decades now"
X Link 2025-09-27T15:58Z [----] followers, [--] engagements
"@giffmana The whole pint of the comic is that for Prof Smith unless it worked it was a unproductive week which is the crux of many students and why they spend uncountable hours in the lab bad professors expect good results all the time"
X Link 2025-09-30T12:54Z [----] followers, [---] engagements
"@Noahpinion Youre making a terrible mistake. This is Buenos Aires"
X Link 2025-10-07T17:05Z [----] followers, [----] engagements
"Si les gust KPop Demon Hunters permtanme introducirlos a la palabra Santa de Macross Frontier"
X Link 2025-10-10T05:00Z [----] followers, [---] engagements
"@Mathinity_ Lol ChatGPT firs stab was with numerical methods"
X Link 2025-10-17T15:17Z [----] followers, [----] engagements
"There are a bazillion papers on crater detection yet no one shares their code Space Science certainly has terrible practices"
X Link 2015-12-08T17:44Z [----] followers, [--] engagements
"Some business are just out there to take advantage of people. charges you [--] USD to check if you need a Visa. http://travisa.com http://travisa.com"
X Link 2016-09-02T00:30Z [----] followers, [--] engagements
Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing
/creator/twitter::leonpalafox