Camilo

One of the most jarring things about current AI is its lack of introspection ability and metacognition. It doesn't know what it doesn't know, how it knows, or how it could find out. It's a one-way system.

One of the many reasons we need real competition. Two large labs is not enough and we need to ensure we have competitive open models.

cron jobs are going through the most epic rebrand

@dvassallo I don't applaud what he did, but good CPAs are worth their time and expertise. Delegating something so delicate to a dumb LLM loop may cost you dearly

This guy set up an alert for my name. He's sitting there waiting for me to get audited so he can celebrate. That's how threatened these people are. The CPA crowd on X might not be representative of all CPAs but holy shit the ones I interact with here seem to think they're brain surgeons. You are putting numbers in boxes and adding them up. You collect receipts and documents from people, you type them into the right fields in the right forms, and you make sure the math works out. Yes there is real value in knowing a lot of tax situations. Years of pattern recognition, knowing which deductions hold up, knowing when something smells wrong. I'm not dismissing that. That part is genuinely valuable. But that part is like 10% of the work. The other 90% is tedious data aggregation into the correct forms and if you cannot see that this is incredibly automatable then I don't know what to tell you. The smart accountants get this. They'll adapt. They'll use the new tools to apply their actual judgment across way more clients with way less grunt work. They'll make more money than they do now. Good for them. But people like this fellow who are fantasizing about me in tax court because I dared to suggest that maybe a human being can do their own taxes with AI assistance. Not cooked. Doomed. "You couldn't possibly understand what I do" is the last thing every automatable profession says before it gets automated.

We software / IT people have a lot to learn from the airline industry. Lesson 1: Male-bro-blame culture won't make your field safer. It will just lead to more cover ups. Lesson 2: Safety first means transparency trumps blame. Focus on outcomes and root causes, not people. Use checklists (& automated checks). Lesson 3: Responsibility and accountability in a complex system is not easy to attribute. Let the process work it out, not the team. Every insight counts, not who reported it.

I'm not very happy with the code quality and I think agents bloat abstractions, have poor code aesthetics, are very prone to copy pasting code blocks and it's a mess, but at this point I stopped fighting it too hard and just moved on. The agents do not listen to my instructions in the AGENTS.md files. E.g. just as one example, no matter how many times I say something like: "Every line of code should do exactly one thing and use intermediate variables as a form of documentation" They will still "multitask" and create complex constructs where one line of code calls 2 functions and then indexes an array with the result. I think in principle I could use hooks or slash commands to clean this up but at some point just a shrug is easier. Yes I think LLM as a judge for soft rewards is in principle and long term slightly problematic (due to goodharting concerns), but in practice and for now I don't think we've picked the low hanging fruit yet here.

por si os encontráis con el término "deuda cognitiva" y os da pereza leer sobre ello: es cuando has vibe-codeado el proyecto entero sin siquiera entender las decisiones estructurales que gepeto ha metido. "¿es malo?" si. "¿por que?" porque cuando hay un error (y gepeto no sabe solucionarlo), no sabes ni por donde te está soplando el viento. "¿es diferente a la deuda técnica?" sip. la deuda técnica es "se que este código o esta decisión que estoy tomando es mala y lo tengo en cuenta para el futuro". La "deuda cognitiva" es "no se lo que estoy haciendo, no tengo ni idea del estado del proyecto, yo solo hablo con gepeto y aparecen cosas en la pantalla, le doy que 'si' a todo y rezo para que funcione". "¿pero esto no sería un problema si gepeto pudiese resolver el 100% de los problemas, no?" correcto. de la misma manera que no tendríamos colas por las mañanas en el bar para pedir un café si existiesen las vacas esféricas que diesen café en vez de leche. "sigo sin entenderlo" tranquilo, ya lo entenderas ;)

You know what’s more sobering? Amazon *already* has some of the most stringent/sophisticated CI/CD game in the business. See x.com/copyconstruct/… Most embracing “vibe coding” don’t have the guardrails Amazon already does and are going to relearn reliability first principles.

@noampomsky people drink and smoke way less

It seems pretty undeniable that people in their 30s/40s today look way younger than people the same age did even 20 years ago. Why is this happening, is it just better cosmetic treatments? What are the second order social effects?

@DamianCatanzaro especulación, ninguna es profitable

La lectura que me pasa con esto es: los laboratorios de AI tienen esas valuaciones GIGANTES porque aún la adopción no es tal pero la herramienta es excelente o esas valuaciones están mal y no correlaciona con el uso actual?

Lo que se habla en tu círculo de las redes sociales y el “para ti” de Twitter es solo un pequeño mundo de lo que realmente está pasando ahí afuera. En cuanto hablan con gente fuera tech/trabajos parecidos la gran mayoría no usa AI o usa ChatGPT gratis y se queja que anda mal porque no saben preguntarle lo que necesitan hacer o que herramienta usar. Aún estamos muy early aunque parezca lo contrario por el ruido que hay a su alrededor.

i’m tired of the learned helplessness in general! no one can do anything anymore. you can’t cook, can’t read, can’t go out, can’t make friends… stop making excuses and start trying! “but capitalism!” ok sure but fight back. reclaim even the tiniest portion of your life, my god!

@PatoAviador ¿cómo son los sistemas electrónicos para la navegación en tierra que podrían prevenir esto? ¿se usan? ¿hay estándares?

Incidente serio en Bruselas Un A320 de Scandinavian Airlines (SAS) con 141 personas a bordo se ve obligado a cancelar el despegue, ya a alta velocidad, al darse cuenta los pilotos de que no estaban en la pista activa, sino en una calle de rodaje. El suceso ocurrió ayer, 5 de febrero, cuando el vuelo 2590 de SAS, un Airbus A320-251N matrícula SE-ROM que cubría la ruta Bruselas a Copenhague, aceleró hasta aproximadamente 117 nudos (unos 217 km/h) en la calle de rodaje. Al parecer, los pilotos confundieron la taxiway E1, paralela a la pista 07R, con la pista asignada, alinearon el avión allí y comenzaron la carrera de despegue. Afortunadamente, la tripulación detectó el error antes de V1 y canceló el despegue (rejected takeoff), pudienti frenar a tiempo. La aeronave se detuvo al final de la calle de rodaje, parcialmente en la hierba y justo antes de una zona de almacenamiento de combustible (fuel farm), evitando una potencial catástrofe. Las condiciones meteorológicas en el Aeropuerto de Bruselas el 5 de febrero de 2026 por la tarde/noche eran buenas: visibilidad excelente (más de 10 km, CAVOK en los METAR), cielo despejado o con pocas nubes altas, y sin lluvia fuerte ni niebla que afectara las operaciones. No se han registrado heridos entre pasajeros ni tripulación. La autoridad aeronáutica belga ha anunciado la apertura de una investigación para determinar lo ocurrido. Fuente: @AviationSafety 👉 aviation-safety.net/wikibase/566414 📸 Imagen del avión involucrado. Fuente: planespotters.net/photo/1846906/…

Fred Brooks in 1999: «Well over half of the time you spend working on a project (on the order of 70 percent) is spent thinking, and no tool, no matter how advanced, can think for you. Consequently, even if a tool did everything except the thinking for you—if it wrote 100 percent of the code, wrote 100 percent of the documentation, did 100 percent of the testing, burned the CD-ROMs, put them in boxes, and mailed them to your customers—the best you could hope for would be a 30 percent improvement in productivity. In order to do better than that, you have to change the way you think.» How much of your work can be outsourced to AI? It is a test. It tells you much about what you have been doing. I write code almost every day. I am also an early adopter of Copilot (AI) to help me code. I probably started using it nearly when it came out. It is great because it helps me do routine work much faster. It is not changing what I do very much. Why not? Because what I did was never, primarily, typing code. I mostly do R&D. It is a fancy way of saying that I mostly try to do “new things”, develop ideas that have not been fully worked out in the past. Even when I use well-understood ideas, my work has to do with testing ideas carefully in practice. It is like my posts here on X. I try to contribute with my original voice. I am enough of a programmer that I could hook up some AI to my account and have it generate daily posts. I could probably do it well enough that a lot of people would be fooled… they would think it is me typing. But that would defeat the purpose. I am trying to push the envelope, to go further than others have gone. It is not, in the least, that I think that I am smarter than others… it is that the value of the information you produce is directly related to its novelty. Repeating what everyone is saying contributes little value. Here is how I imagine software is going to go from now on. You are building your software. There is a well-understood idea that you did not implement for lack of time or lack of people. You go ask AI to implement it, and it will do so quickly. The idea is well understood and the AI has plenty of related training material and references. Done. You, as the developer, have not done nothing even then. You made the choice to adopt this idea, this new technique. Was this a good idea in your context? It is up to you. You take the responsibility. In effect, the “me too” angle in software is a solved problem. You can quickly copy or emulate well-understood ideas. This should impact the software industry in interesting ways. Many software companies have mostly boring code. They have a large stack implementing mostly routine, well-understood ideas. In at least some cases, they should expect to face new and previously unexpected competition. A small team with enough AI power can now reproduce most of their stack. And that's a good thing: innovate or die. But what I also predict is that people who have an “original voice”, people who never just coded what everyone coded, people who went further… these people will do better than ever. For one thing, they will be freed from doing so much routine work. They will have fewer boring coworkers and thus fewer boring meetings. Thus, I predict, people who do R&D will do well. We are not yet seeing the benefits. We don’t see this R&D creating new tangible results… but I think that the “AI wave” will eventually reconfigure enough of the economy, and enough people will figure out the new game, and then we will see something resembling a golden age for innovation. Maybe. I can’t predict the future. But that’s what I imagine.

@lunexb @vhurdaneta @franros89 las tarjetas de bancos distintos son manejadas por el mismo procesador de tarjetas que es el puente con la marca internacional. Sí o sí es un leak o una falla de sistemas de ahí

@camilojd @vhurdaneta @franros89 Fueron dos bancos diferentes, ambos solamente con la tarjeta en Apple Pay ni existe física

No hay caso, por más que uses Apple Pay, los brasileños son expertos en clonar tarjetas. Pagué con NFC en un restaurante del aeropuerto de São Paulo hace 2 meses y hoy empezaron a entrarme montón de compras en una tienda que por suerte el banco me bloqueó).

@lunexb @vhurdaneta @franros89 es completamente imposible. Leak del banco.

@camilojd @vhurdaneta @franros89 Me pasó con dos tarjetas que solo uso con Apple Pay

@vhurdaneta @franros89 me refiero a Apple Pay, está diseñado para que ese ataque sea imposible

@franros89 @camilojd Nfc de tarjeta creo que si se puede, es mejor usar nfc a través de apple pay, mucho mas seguro

@franros89 es como te digo. Demasiadas otras posibilidades hay, inclusive de un leak del banco etc

@camilojd Es lo que siempre pensé yo, pero al día siguiente de pagar en el aeropuerto empezaron a caer compras, y la física está siempre en casa.

There are stuff in real life that cannot be transmitted in books and stuff in books you can never pick up in real life. The disease is the inclination to learn from books what you must learn in real life and attempts to reinvent from experience what can only be found in books.

Working at a company where half the people are based in Germany 🇩🇪 and the other half are based in the US 🇺🇸 really gives you a perspective on how both of these people think and work. Germans are super high maintenance to work with, every new idea needs to come w a set of predefined rules and structure before action is taken meanwhile with Americans you give them some crumbs and as long as the idea feels and seems good they will jump into it and solve problems as they go. Germans think the “American way” is inefficient or careless but the reality is that life is rarely predictable and often rewards action and speed over cautiousness or fear of failure. I think Americans are right about this, risk is something to be managed in motion. Action creates information. Yes, the cost might be rework and sometimes chaos but the upside is momentum and learning.