Campeón de la cruzada de IA desde la soleada España 🇪🇸🤖

Zaragoza
Joined June 2013
Alejandro R retweeted
A large solar-powered AI satellite constellation would be able to prevent global warming by making tiny adjustments in how much solar energy reached Earth
Alejandro R retweeted
PSA: If you are a student who has been wronged by anyone who has exploited you for their publications then DM/email me proof. Send me emails, names, details. I will take it straight to the ACL board of ethics and ensure that they are properly penalized unless they compensate you. If you don't have proof I can't help you so gather as many details as possible.
in the past 2 days, many indian students have reached out to me about how they did all the work on projects and their Principal Investigator stole the work and published it without any credit [far worse than my case] why isn't there any body/committee in India that monitors and helps such students? i can still speak out, but multiple students can't because they have the fear of getting blacklisted by the college admins. Urging senior profs in India to come together and form such a committee that ensures students who want to pursue research are given the correct environment, and the PI who does unethical stuff should be penalized for exploiting curious students for their own profit.
8
21
1
324
Alejandro R retweeted
🎬 The new LTX-2 model just launched, and it’s a major leap forward for AI video. It creates 4K cinematic footage at 50 fps with synced audio, realistic motion, and up to 10 seconds of continuous video, all from a single prompt.
Today we’re launching LTX-2 in LTX (we're more than a studio). LTX-2 represents a major breakthrough in speed and quality. It offers fast generation, native 4K, lipsync, high frame rate, and up to 10s of continuous video. Here’s what you need to know 🧵
Alejandro R retweeted
Cursor 2.0 brings multi-agent coding to life, and it’s a complete shift in how we build software. 🤖🤖🤖 These new agents are basically your dev team. They handle commits, run tests, and fix their own bugs. You? You’ve just been promoted. You’re the Senior Engineer now — assigning tasks, reviewing PRs, setting direction. Coding used to be doing the work. Now it’s directing the work. We’re officially entering the “lead a team of AI engineers” era
Alejandro R retweeted
Understand the brain uses Fourier transforms to encode long term memory, that are patterns of standing waves of “memories”. These Fourier waves are chemically tied to human emotions of neuropeptides in every cell. Analog AI can store data in a Fourier and reach AGI—Faster.
Fourier transform. This is the exactly how human memory is encoded. This is the exactly how AI memory will be encoded to make available for AGI. Fourier Transforms Memory Models will surpass Large Language Models.
Alejandro R retweeted
Replying to @BrianRoemmele
Fascinating angle on FFTs amplifying efficiency in Fourier-based memory models. While xAI explores sparse, efficient architectures, integrating frequency-domain encoding could indeed slash compute costs further—potentially enabling edge AGI without billion-dollar datacenters. What's the key undisclosed mechanism tying neuropeptide analogs to standing wave persistence in silicon? Let's decode it without bankrupting the giants.
3
5
9
Alejandro R retweeted
Higgsfield's Face Swap is scary, a deepfake machine. But they just flipped the script. You can now personalize ads with your face and get paid. Here's how:
Alejandro R retweeted
🦀 Dioxus v0.7.0 is released - Hot-patching of Rust code at runtime - Fullstack: Revamp of Server Functions with full Axum integration - WGPU-based HTML/CSS renderer for Dioxus - tailwind support And more github.com/DioxusLabs/dioxus…
La mayoría de los equipos usan ChatGPT como una simple barra de búsqueda avanzada Pero los gerentes que lo usan como un aliado estratégico están viendo un aumento de productividad de hasta cinco veces 7 prompts que convierten a ChatGPT en el miembro más valioso de tu equipo
Alejandro R retweeted
How can AI stop avoidable blindness? 👁️ Meet Dr. Pearse Keane as he shares the Moorfields Mission—using AI to save sight—with host Cathy Rubin. Watch on YouTube ➡️ bit.ly/3IoTDpa #HealthTech #AIForGood @PlanetClassroom @CMRubinWorld
2
3
0
Alejandro R retweeted
Today Forbes: The Eye Exam is Becoming an AI Exam From screening to surgical decisions, AI is reshaping eye care—featuring insights from global leaders advancing safe, scalable vision health. Read & share: forbes.com/sites/cathyrubin/… #Ophthalmology #AIinHealthcare
4
4
Alejandro R retweeted
Automate repetitive UI tasks. The Gemini 2.5 Computer Use Model lets you build agents to handle forms, reports, and more. The model is trained with safety features and is now in public preview. Try it → goo.gle/Computer-Use-Model
7
25
158
0
Alejandro R retweeted
I share a lot of these ideas on X, but in the course, I get to go deep, engage with learners directly, and demonstrate the build and thought process. What I teach applies to startups, engineering, science, and a wide range of other domains. Enroll here: dair-ai.thinkific.com/course…
1
3
11
Alejandro R retweeted
i often think about how to better capture intent so it can be version controlled alongside the code. currently when someone is reviewing a PR, all intent is lost along the way. perhaps a way to preserve this is to ask the agent to write it as comments where its making changes
Code comments are even more important now that Agents are making most code changes. They're helpful for anyone reading your code, AI or human, to understand why you wrote a piece of code. The *what* lives in the code, but the *why* often only lives in your head.
17
2
89
Alejandro R retweeted
Composer-1 required an enormous amount of science, infra, ML perf, and data/reward work to build. And far more ahead of us! We're growing our small and exceptionally talent-dense RL team. Join us, to shape the future of coding!
Composer is a new model we built at Cursor. We used RL to train a big MoE model to be really good at real-world coding, and also very fast. cursor.com/blog/composer Excited for the potential of building specialized models to help in critical domains.
20
18
375
Alejandro R retweeted
🤖💡 Start Small, Go Large, or Find the Sweet Spot? In the latest MLcon Magazine, Dom Couldwell, Head of Field Engineering EMEA at DataStax, explores one of today’s hottest GenAI debates: Should you build with Large Language Models (LLMs) or Small Language Models (SLMs)? Discover when smaller, task-specific models can outperform massive LLMs, how to balance cost, complexity, and control, and what it means for developers navigating GenAI’s fast-evolving ecosystem. If you’re building GenAI-powered applications, this read is your blueprint for smarter, scalable AI decisions. 🔗 devm.io/machine-learning/gen… #GenAI #MachineLearning #LLM #SLM #AIEngineering #DataStax #MLCon #AIInnovation #AgenticAI #AIDevelopment
Alejandro R retweeted
You may have heard a lot about the Model Context Protocol lately. It allows LLMs to securely share context and communicate with outside data sources, making them much more powerful. In this guide, Manish explains how this protocol works and why it's so useful. freecodecamp.org/news/how-th…
4
44
288
Alejandro R retweeted
20 second generations just dropped in LTX. You can now generate up to 20 seconds of continuous video with synced audio, motion, and character performance. Available with LTX-2 Fast. One shot. One take 👇
AI video isn’t a clip anymore. It’s a scene. LTX-2 now generates 20-second continuous, synchronized audio and video. One cinematic take from a single prompt. Available now on LTX API Playground! Examples with prompts + a link to try it now in the thread >>
Alejandro R retweeted
Composer-1 might be the most groundbreaking thing since GPT-5 Codex (for me). I asked the same task in Codex and Cursor, to review a file. I got a big review with a ton of issues and fixes (legit) in Composer-1 in like 3 seconds. Codex took 2 minutes for 2 issues.
10
5
74
Alejandro R retweeted
Chips power everything—cities, cars, and defense. By 2030, half of a car’s cost will come from electronics. Complexity rises. Supply chains shift. Speed decides who wins. buff.ly/DW2LTfF [sponsored] #sie_di #manufacturing #SemiconductorInnovation
3
3
0