@11krls

Chemisty & Physics | Machine Learning at @acellera

 L'Ampolla, España
11carlesnavarro.github.io/
Joined October 2013
  • Tweets1.6K
  • Following550
  • Followers249
63Media
There is only one basic human right, the right to do as you damn well please. And with it comes the only basic human duty, the duty to take the consequences.
1
9
I just wanna talk to my computer and there's nothing that just works standalon. I'll need to build my own
34
Guys is not intelligence is distribution sampling
24
Writing is a great exercise, it forces you to get your ideas out of your head and see how bad they are.
26
Look, I want a personal agent but with access to my local data, not in a separate cloud with a computer I don't have access to. It needs to be an extension of me, I want to talk about what I am doing on my computer, my phone, and his computer.
Claude
@claudeaiOct 21
We’re opening a research preview of Claude Duet, a new app where Claude works as a proactive agent. It comes to you with what needs your attention, completes tasks you ask of it, keeps working when you step away, and checks with you before it acts on your behalf.
46
Using opus to learn through videos is the next step after html artifacts. I just have on-demand youtube videos
1
105
Carles Navarro retweeted
PlayMolecule@PlayMoleculeOct 15
Need a new scaffold for your target protein pocket? In this one-minute demo, see how PlayMolecule AI takes on the challenge. Explore the platform: playmolecule.ai/
1
53
Carles Navarro retweeted
François Chollet
@fcholletOct 14
String theory is what happens when an intelligent system is starved of sensory input and starts hallucinating. Intelligence needs grounding. For this reason, intelligence is practically never the bottleneck to solving impactful problems. The bottleneck is iterated contact with reality.
766
1,307
334
12,482
5,508,405
Carles Navarro retweeted
François Chollet
@fcholletOct 14
The ideal outcome for science isn't a couple of labs using proprietary models and millions of dollars to crack trophy problems, it's to have millions of AI-aided human scientists who follow their curiosity and become way more ambitious
160
344
56
2,645
91,116
shouldn't ASI need less PRs? why would a super intelligent system need to change its code 30k times per day? wasn't the first change super-intelligent enough?
Anthropic
@AnthropicAIOct 13
New on our Engineering blog: Creating in-house source control. In May, we merged around 2,500 code changes per day. In September, we merged more than 30,000 per day. As a result, we became bottlenecked by developer infrastructure built for human (rather than agent) scale. Here, we share how we built a new version control system—and fully migrated our monorepo from git—in five months: anthropic.com/engineering/re…

Redesigning source control at Anthropic

As our merges per day grew more than tenfold, source control became our biggest development bottleneck. Here, we share how we migrated to an in-house version control system in five months.

anthropic.com
22
To me, something being “solved” means it can be produced automatically, without any human work beyond asking for it. I asked for a quantum-level molecular simulation engine that could predict the exact binding energy of a protein–ligand system in under two days. It didn’t happen. I thought software was solved, though…
34
The only coding harness you need
24
Carles Navarro retweeted
Nikita Bier
@nikitabierOct 12
Technical moats are mostly over at this point. If you’re building something that is meant to be durable, the only moats now are: • Proprietary data • Network effects • A physical component
1,151
2,084
902
25,783
3,318,999
Math seems to be interpolation so
Epoch AI
@EpochAIResearchOct 7
AI developers aim to create an automated AI researcher. How close are they? To find out, we built InnovationEval, which tests whether AI can produce post-training innovations comparable in magnitude to a recently published advance. So far, agents’ results are underwhelming.
31
Speak to a Protein has been accepted at JCIM! We started building our first AI co-scientist over a year ago, which led to this paper. It took careful thinking about how to give agents the freedom to do science, make their working environment secure, and build a shared workspace where scientists could work alongside them. Since then, a lot has changed in agentic AI. In PlayMoleculeAI we can now run teams of agents on drug discovery campaigns for days or even weeks, generating hypotheses, testing them through computational chemistry jobs on cloud CPUs and GPUs, and proposing new molecules for synthesis. We see this as the future of in silico molecular design: scientists working with teams of agents that can run research campaigns in a data center.
1
2
60
Paper: doi.org/10.1021/acs.jcim.6c0… Blog: acellera.com/blog/how-we-cre…

Speak to a Protein: An Interactive Multimodal Co-Scientist for Protein Analysis

Abstract. Building a working mental model of a protein typically requires weeks of reading, cross-referencing crystal and predicted structures, and inspect

pubs.acs.org
11
opus 5.5 haciendo excalidraw-like diagrams es lo unico que necesito
25

What Happens When We Stop Needing Each Other?

Article · @11krls
38
fuck now I just talk to agents but if I am not wearing headphones I need to stop videos every single time
30
Carles Navarro retweeted
Vicent Martí
@vmgSep 21
This is nothing. There's no frontier AI development in this country. The very few Spanish engineers working on frontier AI are working for American companies. It is literally impossible to develop competitive AI in the socialist hellscape that this guy has thrust upon us.
Pedro Sánchez
@sanchezcastejonSep 21
Hoy presentamos el Plan IA360 para un despliegue responsable de la Inteligencia Artificial. Un plan con acciones concretas a desarrollar en los proximos 12 meses para que España aproveche las oportunidades que la IA ofrece a nuestra economía. Poniendo en el centro la seguridad ciudadana, la confianza y la protección de los más vulnerables, empezando por los menores.
35
126
6
1,150
88,413
Load more