Juanmi
banner
juanmirod.bsky.social
Juanmi
@juanmirod.bsky.social
Reposted by Juanmi
The best approach I know of for reducing the risk here is to make sure that any credentials that are visible to coding agents - like AWS keys - are tied to non-production accounts with strict spending limits

That way if the credentials are stolen the blast radius is limited
November 25, 2025 at 9:00 PM
Every AI Breakthrough Shifts the Goalposts of Artificial General Intelligence | Scientific American share.google/eMLlkdZXWuCV...
Each Time AI Gets Smarter, We Change the Definition of Intelligence
As AI systems exceed one benchmark after another, our standards for “humanlike intelligence” keep evolving
share.google
November 24, 2025 at 6:08 PM
I thought: "It would be nice to visualize 6D vectors using 3 dimensions for the position and 3 for the color" so I asked Gemini to generate a 6d vector visualizer for me, and it did it, one shot: gemini.google.com/share/2ba656...
November 21, 2025 at 9:38 PM
The new capabilities of nano banana are amazing, has someone managed to make it use a concrete color palette? I tried with an image and with hex codes, but both times it just used "similar" colors.
Nano banana Pro: “i need a flowchart for how to toast bread, make it as wacky and over the top and complicated as possible.“

Not absolutely perfect, but I can’t believe how much there is a coherent through-line, how clear the text is, and also parts of it are actually funny?
November 21, 2025 at 8:45 AM
Reposted by Juanmi
November 19, 2025 at 3:36 PM
Reposted by Juanmi
Está claro, lo sorprendente del caso es que sean magistrados del Tribunal Supremo los que se dediquen a hacer lawfare de manera tan descarada. Pero los que fuimos siguiendo los juicios del Procés ya estamos curados de espanto. También los de Alsasua, las 6 de La Suiza... Y un largo etcétera.
November 20, 2025 at 8:10 PM
Elon creó Grok con la única motivación de que pudiera decir estás mierdas a millones de personas y hacer propaganda nazi a gran escala, cualquier otra motivación era secundaria y lo ha dicho y demostrado en numerosas ocasiones. Quedarse a leer esa propaganda es contribuir con tu tiempo y tu atención
Calling on the Irish government and state agencies to immediately close their X accounts. Continued use of a fascist agitator's social media site is both incompatible with democratic values and immoral.
November 20, 2025 at 5:53 PM
Esto de no saber si estallará antes la burbuja de la vivienda o la de la IA es un sinvivir.
November 19, 2025 at 8:54 PM
November 17, 2025 at 10:09 PM
El bullshit-meter de Sabina nunca falla 😁 www.youtube.com/watch?v=k6Ad... (Disclaimer, no creo que vivamos en una simulación, como no creo en Santa ni en los unicornios, pero no han "demostrado" una mierda)
www.youtube.com
November 14, 2025 at 6:09 PM
Reposted by Juanmi
Some pretty eye-opening data on the effect of AI coding.

When Cursor added agentic coding in 2024, adopters produced 39% more code merges, with no sign of a decrease in quality (revert rates were the same, bugs dropped) and no sign that the scope of the work shrank. papers.ssrn.com/sol3/papers....
November 13, 2025 at 5:18 AM
Reposted by Juanmi
📅 Pon el 11 de noviembre en tu calendario.

Organizamos 🤖 Algoritmo público, código abierto, una jornada para abrir las cajas negras de las administraciones.

Hablaremos de cómo auditar sistemas que deciden sobre nuestras vidas.

📍 La Corrala, Madrid.

👉 Apúntate gratis hoy:
Algoritmo público, código abierto | El derecho a destripar las cajas negras de las administraciones
El 11 de noviembre Civio organizará un evento que reunirá a expertos, periodistas, activistas e instituciones que charlarán sobre el uso de los algoritmos en las administraciones públicas y, sobre…
civio.es
October 2, 2025 at 7:42 AM
Reposted by Juanmi
OpenAI partially released a new model yesterday called GPT-5-Codex-Mini

No API access yet, but I did some truly horrible things to their Codex CLI app to get it to spit out this SVG of a pelican riding a bicycle
November 9, 2025 at 3:37 AM
Reposted by Juanmi
I dug into progressive image rendering and found a bunch of common assumptions aren't quite true.

➡️ JPEG & WebP progressive-renders worse in Safari
➡️ AVIF _does_ support progressive
➡️ But JPEG XL in Safari doesn't
➡️ JPEG XL decodes much slower than AVIF
⬇️ and more

jakearchibald.com/2025/present...
The present and potential future of progressive image rendering
Exploring progressive image rendering across JPEG, PNG, WebP, AVIF, and JPEG XL.
jakearchibald.com
October 16, 2025 at 10:16 AM
Reposted by Juanmi
“How can we make it so that all human beings have enough?”

Superhuman intelligence: well the people who have way too much need to give something up

Techbros: no no no CHANGE THE ALGORITHM
November 7, 2025 at 3:05 PM
Lo de los robots sirvientes que en realidad son teleoperados por alguien con aún peores condiciones laborales que las trabajadoras de un hotel no me lo vi venir la verdad.
November 6, 2025 at 5:02 PM
Así que, a ver qué me entere, OpenAi está pidiendo que el gobierno les "avale" la deuda para que así les pongan menos intereses, porque se han comprometido a gastarse 1 billón de dolares (en español) en servidores...
November 6, 2025 at 4:54 PM
November 6, 2025 at 4:47 PM
Reposted by Juanmi
“There's no ability to allow Palestinians to go back to any semblance of a dignified life” in Gaza, says Tareq Baconi, Palestinian analyst and writer. “The narrative is of a ceasefire. The reality is of the continuation of the genocide.”
November 6, 2025 at 3:16 PM
Reposted by Juanmi
The Horror...
The Horror...
@ceosgalegos.bsky.social
October 29, 2025 at 5:17 PM
Reposted by Juanmi
Penpot's new MCP Server — open, self-hostable & LLM-agnostic — takes AI workflows far beyond “design-to-code.”

Try it, break it, build with it → we need your help to stress-test and share new ideas!
community.penpot.app/t/penpot-mcp...
Penpot MCP Server showcase + ask for help
Only three weeks ago, during Penpot Fest 2025, we unveiled the official MCP Server for Penpot (Github repo and 6-min presentation). Unlike other community efforts (kudos to them!), we decided to expos...
community.penpot.app
October 30, 2025 at 6:13 PM
Reposted by Juanmi
From this new post by OpenAI: 0.15% of users (something like 9M people given public numbers) show signs of suicidal intent in their ChatGPT chats each week

But there seems to be progress in making ChatGPT respond appropriately to mental health issues. openai.com/index/streng...
October 28, 2025 at 4:40 AM
Reposted by Juanmi
If you demand that people control something they can't control, then expect them to control something else. If the mandate is "No Bugs!" without any social or technical change, then expect a response like "no *reported* bugs".
October 22, 2025 at 5:13 PM
Todos los años la misma discusión y la misma pantomima
🔴 ÚLTIMA HORA | El Gobierno propone acabar con el cambio de hora estacional en la UE desde 2026 social.elpais.com/wdvmq5
October 20, 2025 at 4:54 PM
El nivel de seguidismo a cualquier salida de tono o mamarrachez que dice el rey naranja es increíble, imagino que todos piensan "ande yo caliente..." O directamente están de acuerdo en que la estrategia sea generar confusión mientras ellos expolian lo público a manos llenas...
October 20, 2025 at 4:26 PM