But no one besides Novartis& Lilly etc. has access to the ligand structure prediction. All the data on ligand binding in the paper is irreproducible and should not have been published. As Reviewer #3 for @nature I recommended taking it out and saving it for promotional material.

Companies like Deepmind do not have any requirements to make their research available. I signed the letter below because they chose to publish their work and different standards were applied to the Alphafold 3 publication as those applied to others.

¿Por qué el nuevo AlphaFold3 DeepMind es sólo accesible via web, con límite de peticiones, lo cual impide hacer búsquedas de nuevos fármacos a los científicos de la comunidad? Porque dichas búsquedas ya las esta haciendo una spinoff de ellos mismos con farmacéuticas 👇

nature.com/articles/d4158…
'Access to the AF3 server ... is limited. Scientists are currently restricted to 10 predictions a day, and it is not possible to obtain structures of proteins bound to possible drugs' or to do systematic studies.
Not good for Deepmind reputation.

Heading home from #bog24 on the overday flight (sorry to miss the last session) and here are some thoughts from this Biology of Genomes

AlfonsoValencia [email protected] En todo caso sería el plus de desinformación conseguido con la IA, que de momento no parecen necesitar ni Trump ni la extrema derecha para elaborar sus mensajes …

AlfonsoValencia [email protected] Las IA diseñadas para juegos donde el engaño forma parte como el poker o las que se ponen objetivos sin restricciones morales pues engañan. Es un artículo tendencioso.

Aina BSC-CNS TIC Catalunya Una experiència súper interessant, gràcies per la visita

Las inteligencias algorítmicas se hacen espacio en la ecología de inteligencias hasta ahora biológicas, recibiendo, emitiendo y procesando información de formas extremadamente complejas.

Curious about the use of #DeepLearning for genomic sequences to study transcription? Check out our Annual Reviews review

Las IAs ya han aprendido a engañar y manipular a los humanos lavanguardia.com/vida/20240511/…
Considerar los sistemas de IA que pueden mentir (todos?) de alto riesgo según la IA act?

#OpenSoftwareAcceleratesScience

I feel like playing with DALLE... The models look great but when you look closer some things are new and impressive, but some things seem hallucinated. Care and experimental validation thus as always required!
x.com/jankosinski/st…
x.com/jankosinski/st…

🥉 ¡Cataluña es 3⃣a en la UE en n.º de publicaciones sobre terapias avanzadas!

El Hospital Clínic, @ValldHebron, Hosp. Germans Trias, Sant Pau - Campus Salut Barcelona, Hospital Sant Joan de Déu Barcelona ES, el Banc de Sang i Teixits, y Leitat, entre otros, han contribuido significativamente.

🔗 report.biocat.cat

Imperial College estimated that together, Covid vaccines saved about 20 million lives by the end of 2021.

A further study by vaccine type, estimated that the AZ vaccine saved over 6 million lived by the end of 2021. 6/9

AlfonsoValencia [email protected] Michael Hoffman @michaelhoffman.bsky.social Roland Dunbrack 🏳️‍🌈 @rolanddunbrack.bsky.social Mick W@tson ↙️ nature Google DeepMind elpais Demis Hassabis totally agree with you and quite a few others key scientists- either you are fully opened or you keep it as a trade secret

El Dúo Termodinámico. 🤣🤣🤣

Telegraph: Microsoft reportedly planning 'Stargate', a $100billion supercomputer to be powered by several nuclear plants

The era of the great uranium brains approaches.

This is the perfect justification for closed source AI. We will see many more “As part of our commitment to releasing our research safely and responsibly, we will not be sharing model weights.” from now on.

Michael J. Sandel: “Me resisto a la tendencia de ver la tecnología como una fuerza autónoma que no podemos controlar” elpais.com/sociedad/2024-… vía EL PAÍS por Pablo Guimón