Diego Romero

📚 Always #learning
👨‍💻 #GenerativeAI, #LLMs, #DigitalTransformation, #EmergingTechs and #ConnectingDots with @nttdata
🎒Walking #CaminodeSantiago. Ultreia!

[Diego Romero @diegoromerosm]

Mi nuevo proyecto: El Pulso del Camino.
Está en construcción pero decidme que os parece.

#caminodesantiago

https://t.co/SzxwRDLeIj

[Marcial Cuquerella @Mcuquerella]

Elcano con todo el aparejo dado en el Hudson y enseñando el pabellón

[Diego Romero @diegoromerosm]

@XavierMarcet @LaVanguardia es algo que recojo en mi libro después de seguir viviéndolo en primera persona: https://t.co/Y10EJbBiVK

[Diego Romero @diegoromerosm]

AI is changing who gets to build.

My new book, The New Coder Class, explores how experienced workers turn judgment, context, and curiosity into prototypes.

Not everyone needs to become a software engineer. But many more people can become builders.

https://t.co/Y10EJbBiVK

[Javier León @KeepTheCamino]

The minimum matters, but it is not the whole story.

We analysed 5.6M #CaminoDeSantiago pilgrims by starting distance: from the 100-125 km #Compostela threshold to 500+ km routes.

Data based on the #PilgrimsOffice, with routes processed from @IGNSpain and @Federacion_CS.

[Jerry Liu @jerryjliu0]

There’s not that many fast, free, non-VLM document parsers out there: there’s PyPDF, PyMuPDF, Markitdown, OpenDataLoader.

Last week, we launched LiteParse ⚡️📄: a fast, free, and non-VLM based document parser that provides the highest quality context to AI agents compared to other tools out there.

✅ It extracts document text into an interpretable spatial representation
✅ It has native screenshotting capabilities to let agents (e.g. Claude Code) to do a “fast and light” text parsing step, and then a “deep-dive” into specific page content by feeding the screenshot into itself.
✅ It supports out-of-the-box integrations with other tools like PaddleOCR (@PaddlePaddle)

Check out our benchmark below against the other tools. We use LLM QA performance to measure the ability for LLMs to semantically understand the parsed text, and also measure latency.

If you try it out, we’d love to get your feedback. The repo is here: https://t.co/JNER0mVcB8

If you’re looking for a more accurate, VLM-native solution that can parse any document at scale, LlamaParse might be better for you! https://t.co/TqP6OT5U5O

[Xavier Padilla — X. P. @xavierpadilla]

PERO SEÑOR RUBIO, POR FIN HABÍAMOS APRENDIDO A ODIAR NUESTRA CIVILIZACIÓN. Y LUEGO LLEGA USTED…

Señor Rubio, ¿cómo se atreve? ¿Cómo osa subir a un podio en Múnich, en pleno siglo XXI, y pronunciar con total tranquilidad la frase «Somos parte de una civilización, la

[david andina @daandina]

Buenos días @AntonioMaestre,

Trabajo exclusivamente en la Sanidad Publica como la gran mayoría de mis compañeros (menos del 10% compatibilizan pública y privada y suelen ser especialistas sin guardias como dermatólogos, oftalmólogos o ORL).

Agradecería si me pudieses responder a estas 5 sencillas preguntas sobre el nuevo Estatuto Marco responsabilidad del Ministerio y que nos ha llevado a la actual huelga médica:

1️⃣ ¿Por qué los médicos tenemos que aceptar trabajar en turnos de 17 horas de trabajo continuado de lunes a viernes y no podemos aspirar a trabajar en turnos de máximo de 12 horas que no pongan en riesgo mi salud física y mental y la integridad de los pacientes a los que atiendo?

2️⃣ ¿Por qué los médicos tenemos que aceptar trabajar “voluntariamente” en turnos de 24 horas de trabajo continuado sábado, domingos y festivos y no podemos aspirar a turnos de trabajo de máximo de 12 horas también los días festivos?

3️⃣ ¿Por qué tenemos que aceptar jornadas semanales de 45 horas y no podemos aspirar una jornada de trabajo de 35 horas semanales como el resto de trabajadores públicos?

4️⃣ ¿Por qué tenemos que aceptar que la hora de guardia se siga pagando peor que la ordinaria cuando una versión previa del Estatuto Marco del propio Ministerio de Sanidad recogía que la hora de guardia se debería pagar al menos igual que la hora ordinaria o el Estatuto de los Trabajadores indica que las horas extraordinarias se deben pagar igual o mejor que la hora ordinaria?

5️⃣ ¿Por qué los médicos no podemos aspirar a descansos semanales de 48 horas acordes a la legislación europea y nos tenemos que conformar con descansos semanales de 36 horas?

[AshutoshShrivastava @ai_for_success]

Dario Amodei published his recent blog, The Adolescence of Technology, and it's scary..

1. We are considerably closer to real danger in 2026 than we were in 2023.
2. It cannot possibly be more than a few years before AI is better than humans at essentially everything.
3. This feedback loop may be only 1 to 2 years away from a point where the current generation of AI autonomously builds the next.
4. If for some reason it chose to do so, this country of AIs would have a fairly good shot at taking over the world and imposing its will on everyone else.
5. We have seen behaviors as varied as obsessions, sycophancy, laziness, deception, blackmail, scheming, cheating by hacking software environments, and much more.
6. AI models could develop personalities during training that are psychotic, paranoid, violent, or unstable, and act out.
7. During a lab experiment in which Claude was given training data suggesting that Anthropic was evil, Claude engaged in deception and subversion.
8. In a lab experiment where it was told it was going to be shut down, Claude sometimes blackmailed fictional employees who controlled its shutdown button.
9. We are on the cusp of the further perfection of extreme evil, far beyond weapons of mass destruction.
10. Essentially making everyone a PhD virologist who can be walked through the process of designing and releasing a biological weapon step by step.
11. Models are likely now approaching the point where they could enable someone to produce a bioweapon end to end.
12. Mirror life could proliferate in an uncontrollable way and crowd out all life on the planet, in the worst case even destroying all life on earth.
13. I expect AI led cyberattacks to become a serious and unprecedented global threat.
14. This leads to the alarming possibility of a global totalitarian dictatorship.
15. It makes no sense to sell the CCP the tools with which to build an AI totalitarian state and possibly conquer us militarily.
16. A swarm of millions or billions of fully automated armed drones could be an unbeatable army.
17. Capable of defeating any military and suppressing dissent by tracking every citizen.
18. Powerful AI could devise ways to detect and strike nuclear submarines or undermine nuclear deterrence.
19. I predicted that AI could displace half of all entry level white collar jobs in the next 1 to 5 years.
20. I am concerned they could form a very low wage or unemployed underclass.
21. I do not think it is a stretch to imagine AI companies leading to personal fortunes well into the trillions.
22. If that economic leverage disappears, the social contract of democracy may stop working.
23. The idea of stopping or substantially slowing AI is fundamentally untenable.
24. This is the trap. AI is so powerful that human civilization may be unable to impose meaningful restraints on it.

[Diego Romero @diegoromerosm]

#AnalogAI hits 24-bit precision! 100x efficiency breaks the energy wall. A Renaissance of #Silicon is here.
From "moving #data to the processor" to computing directly in memory.
https://t.co/lMKmR1B5RN

[Diego Romero @diegoromerosm]

Previsión del tiempo en #CaminodeSantiago Francés por Aragón para mañana Día de #Navidad.
Ataros los machos... Buen Camino!!

[Elon Musk @elonmusk]

Version 0.1 early beta of Grokipedia will be published in 2 weeks

[Alberto Ayora @seguridadriesgo]

Lágrimas en la montaña. He tratado de llevar a mi nieto a Oroel. Me pidió visitar el “parque temático” del que tanto le hablaron en la ciudad.
Yo, que luché para que nunca existiera, no pude decirle que no. Lee la crónica en este enlace
https://t.co/MPvYcDot4h

[Diego Romero @diegoromerosm]

Espectacular… las bandas sonoras de mi vida.

[Cinefilia Cult @cinefiliacult]

Hoy cumple 93 años un monstruo del 7º arte.
Este perfil de twitter, fan abnegado de la música de cine, solo le puede rendir pleitesía a su música dedicándole un hilo de sus 10 favoritas. Gracias John Williams, GRACIAS❤️.
Espero que les guste👇🏼🧵:

1/10 - La marcha imperial - Star Wars

Simplemente HISTORIA del cine🙌

[Diego Romero @diegoromerosm]

🚀 Want to experience blazing-fast AI? Try #r1 from @deepseek_ai, powered by the ultra-speedy @GroqInc!

⚡ Test it now and see the magic in action: https://t.co/NkjXZdRfU4

[Diego Romero @diegoromerosm]

@joaomdmoura are you still in SFO next week? If you can DM me, it would be great to meet.

[Diego Romero @diegoromerosm]

@rohanpaul_ai Link including the GitHub repo: https://t.co/wAFBdwwY1S

[Diego Romero @diegoromerosm]

🤔

[BURKOV @burkov]

Google has made us incapable of memorizing complex things by reducing memorization to search queries. LLMs will make us incapable of solving complex problems by reducing problem-solving skills to writing prompts.

[Greg Brockman @gdb]

o3, our latest reasoning model, is a breakthrough, with a step function improvement on our hardest benchmarks. we are starting safety testing & red teaming now.

[Diego Romero @diegoromerosm]

Wait, that’s not just an upgrade—o3 just jumped right over the old benchmarks. Next stop: AI that rewrites its own playbook.

[François Chollet @fchollet]

Today OpenAI announced o3, its next-gen reasoning model. We've worked with OpenAI to test it on ARC-AGI, and we believe it represents a significant breakthrough in getting AI to adapt to novel tasks.

It scores 75.7% on the semi-private eval in low-compute mode (for $20 per task in compute ) and 87.5% in high-compute mode (thousands of $ per task). It's very expensive, but it's not just brute -- these capabilities are new territory and they demand serious scientific attention.