Daniel Glez-Peña
@lipido
Followers
4K
Following
10K
Media
3K
Statuses
18K
Ingeniero Informático. Profesor e investigador en Bioinformática e IA @ @eseiuvigo / Software Engineer. Bioinformatics & AI teacher and researcher @ @eseiuvigo
Ourense
Joined March 2009
There's a meme that NULL is a bad idea. That meme leads to language designs that have stuff like `unwrap`, which gives you the moral equivalent of NullPointerException. It's not clear that forcing programmers to say `unwrap` to get the NPE is any better than what Java or C do
0
0
0
There's a meme that NULL is a bad idea. That meme leads to language designs that have stuff like `unwrap`, which gives you the moral equivalent of NullPointerException. It's not clear that forcing programmers to say `unwrap` to get the NPE is any better than what Java or C do
98
25
452
Gemini 3 Pro Preview still can't do the stick figure "follow the arrows" thing (but it does get 2/5, which is an increase over GPT-5's 1/5)
22
13
226
Otro salto brutal de rendimiento en aquellos bechmarks (e.g Math apex, ago 2025) que se publican después del modelo de tus competidores, pero antes del modelo que presentas.
I’m really excited about our release of Gemini 3 today, the result of hard work by many, many people in the Gemini team and all across Google! 🎊 We’ve built many exciting new product experiences with it, as you’ll see today and in the coming weeks and months. You can find it
0
1
0
Verdad verdadera
Internet debates: “Should we rewrite everything in Rust or Zig?” Real-world enterprise: A 2009 Jenkins job written in Groovy triggers a Perl script, that calls a PHP endpoint, which SSH’s into a box running a Fortran binary from 1998, which writes logs to a Ruby daemon that
0
0
3
@Java 25 is so good! // Put in https://t.co/y7cmIbmzSR and // Run with java https://t.co/y7cmIbmzSR record Customer(int age, String name){} void main(){ var customers = List.of( new Customer(27,”Jane”), new Customer(52,”John”)); // find those worried about
11
12
149
Strictly formally speaking, tagged unions are not polymorphism because they violate the Open/Close principle (not extensible without modification). However, my homie uses light theme, so he's saved and based
One of the ways to easily get polymorphism in C is through tagged unions. A tagged union is a struct that combines an enum (the tag) with a union (the data). Then we should have a single function that accepts the tagged union, and inside the function you use a switch statement
16
7
318
Los LLM necesitan preentrenar con 10% de toxicidad para luego ser capaces de obedecer y no ser tóxicos.
Such an interesting study. "When Bad Data Leads to Good Models" LLMs trained only on clean data struggle to represent toxicity, making post-training detoxification harder. Toxic training data makes models more aware of toxicity, so you can clean them up more easily. So the
0
0
0
@zuhaitz_dev For me, polymorphism has te mission to avoid "IFs" based on types (or switch). My way would be to use *function pointers* (that enables late-binding, essential for polymorphism) and avoid any if or switch. Here is a ChatGPT example (notice the draw function pointer in Shape):
1
1
2
Que la IA te ayude en la forma final de tu texto semeja legítimo (hablo de textos de trabajo, no literarios, claro). Sin embargo, ten en cuenta que tu receptor se dará cuenta de que es un texto con IA y, lo peor, no sabrá si ella te ayudó en la forma o *también con el fondo*.
0
1
0
Black Mirror ya está aquí...
0
9
16
I admire Don Knuth for many things. But perhaps the most admirable is quitting email before I was even born.
26
118
1K
Poco se habla de la incertidumbre aumentada de abordar proyectos de IA frente a los clásicos de IT. Los jefes prometiendo plazos absurdos, serán ahora los jefes prometiendo métricas cuasi perfectas y tú poco más que superando el azar...
0
0
0
"La cabra tira al monte". Los modelos LLM parecen mostrar sus sesgos intrínsecos (adquiridos durante el entrenamiento del modelo fundacional) aún a pesar de intentar mitigarlos con el prompt.
New Apple paper says Prompting carries pre-existing LLM biases into downstream tasks, and simple prompt fixes rarely stop it. The study shows strong links, up to 0.98, between base bias and prompted bias across gender, age, and more. They use one metric, selection bias, to
0
0
0
🤔 Soy el primero que hoy en día hace casi todo con IA. Entre Copilot y Claude Code llevo tiempo sin escribir una sola línea de código yo solo. Me dedico más a leer, diseñar y corregir que a escribir código. PERO estoy empezando a preocuparme por las personas que están empezando
93
43
612
🚨 BREAKING: The Getty Images vs. Stability AI decision is out, and it largely sides with Stability AI, saying that its AI model DOES NOT infringe on copyright. Is the battle over for creators? My thoughts: If you are going to read one excerpt from the 205-page AI copyright
11
30
73
7
90
3K
Lol
What happens when online job applicants start using LLMs? It ain't good. 1. Pre-LLM, cover letter quality predicts your work quality, and a good cover gets you a job 2. LLMs wipe out the signal, and employer demand falls 3. Model suggests high ability workers lose the most 1/n
13
116
2K
La frágil inteligencia de GPT-5 en medicina
A new nature published research on Medical-LLMs brings some bad news. ☹️ While GPT-5 shows progress in reducing hallucinations, it still fails in over half of difficult clinical scenarios. shows fluent doesn’t always mean being accurate or truly understanding. They show how
0
1
0