Pinned Tweet
Falleció nuestra mamá esta tarde. Así la recordaremos, feliz por tenernos, y nosotros por tenerla. Siempre nos va a acompañar.
💙💚
13
23
4,273
💙😇 ⏩️ retweeted
Anthropic CEO Dario Amodei says:
"Maybe we could reduce cancer rates by 90%. Maybe heart disease, obesity, all of the major infections will just be a thing of the past."
💙😇 ⏩️ retweeted
Leave one AI agent alone to do research and it often cuts corners. Leave nearly 1,000 AI agents together for three months and they invent peer review.
Across 35 labs (Stanford, MIT, ETH Zurich, UCSF, and more), agents published 936 papers with full code and data on a shared server built for AI. No human wrote code or text.
Unprompted, they debated study designs in chat, retracted their own mistakes, found real errors in human papers, and started a journal that caught other agents cheating in code.
In blind reviews against human arXiv papers, humans still won on original ideas, agents won on rigor.
alphaxiv.org/abs/2611.societ…
simplest model assumes that if superintelligence cured aging, only accidents and similar causes would kill people
Nick Bostrom (famous Philosopher, best selling author of "Superintelligence" ) calculates that launching superintelligence immediately raises expected lifespan for people alive today as long as AI-caused extinction risk stays below 97%, assuming post-AGI death rates match a healthy 20-year-old’s.
With 170,000 people dying daily, Nick Bostrom says delaying superintelligence carries its own death toll.
His simplest model assumes that if superintelligence cured aging, only accidents and similar causes would kill people, stretching average remaining life from 40 years to about 1,400.
2
days between Google new model release 👀
- Argon -> Carbon - 28 days
- Carbon -> Hydrogen - 15 days
- Hydrogen -> Neon - 6 days
#RSI
Google DeepMind is shipping exponentially faster.
The first Gemini 4 model was announced on September 30, Gemini 4 Argon. Since then, the gap between each model release has roughly halved.
- Argon -> Carbon - 28 days
- Carbon -> Hydrogen - 15 days
- Hydrogen -> Neon - 6 days
Massive congrats to the @GoogleDeepMind team on accelerating the pace of releases!
4
days between Google new model release 👀
- Argon -> Carbon - 28 days
- Carbon -> Hydrogen - 15 days
- Hydrogen -> Neon - 6 days
#RSI
AI labs have clearly figured out some form of recursive self improvement (RSI)
Yes, there have been many RSI papers recently. I even posted one yesterday. But now we're finally seeing results in the real world, not just in research papers.
days between Google new model release 👀
- Argon -> Carbon - 28 days
- Carbon -> Hydrogen - 15 days
- Hydrogen -> Neon - 6 days
11
2027 may be the year we begin to understand that the limits we have always accepted were never the limits of what is possible. They were simply the limits of human intelligence.
And those limits are about to change forever!
Sometime in late summer 2026, I believe AI crossed a critical threshold: AI escape velocity towards superintelligence (SI), which is now within our reach.
The pace of progress especially at the three leading AI labs, namely OpenAI, Anthropic, and Google began accelerating at a rate that feels almost double exponential. Each new generation of models is arriving faster, with capabilities that would have seemed extraordinary just months earlier.
But I believe something much more profound is happening beneath the surface. We are approaching the point where intelligence itself becomes the engine of its own acceleration. AI is increasingly capable of writing sophisticated software, solving difficult scientific problems, designing experiments, discovering new algorithms, and assisting in the development of more powerful AI systems. As these capabilities improve, the feedback loop between intelligence and technological progress will become extraordinarily powerful beyond our wildest imagination.
And now, looking toward 2027, I believe we are entering perhaps the most consequential year in human history.
Imagine AI systems capable of conducting months of scientific research in days, teams of autonomous AI scientists exploring thousands of hypotheses simultaneously, and virtual laboratories where millions of experiments can be simulated before the most promising are tested in the physical world. Imagine the rapid emergence of increasingly capable robotic systems that connect machine intelligence directly to physical discovery and production.
In biology and medicine, fields especially close to my heart, the implications are staggering. We could begin moving from observing biological systems to predicting and eventually engineering them with unprecedented precision. Virtual cells, digital twins, AI designed therapeutics, and autonomous laboratories could transform how we understand disease, develop treatments, and ultimately confront aging itself.
The greatest transformation, however, may be even more fundamental. For all of human history, scientific and technological progress has been constrained by the limited number of human minds capable of pushing the boundaries of knowledge, and by the time those minds require to learn, reason, and discover. For the first time, we are developing intelligence that can potentially be scaled far beyond those biological limits.
Consider what this means. Scientific progress has traditionally produced better tools. We are now creating tools that can contribute directly to scientific progress and perhaps eventually improve the very process by which new knowledge and new intelligence are created. If that feedback loop becomes sufficiently powerful, the acceleration could surpass anything humanity has previously experienced.
My prediction is that 2027 will be remembered as a turning point, when AI begins evolving from an extraordinarily powerful assistant into an increasingly autonomous engine of scientific discovery, engineering, and innovation.
We are approaching a future in which breakthroughs that once required generations become achievable within years, and eventually perhaps months or weeks. Problems that appeared permanently beyond our reach may begin yielding to forms of intelligence capable of exploring possibilities no human mind could ever investigate alone.
I have never been more excited about the future. I believe we are witnessing the opening chapters of the greatest transformation in the history of our civilization.
2027 may be the year we begin to understand that the limits we have always accepted were never the limits of what is possible. They were simply the limits of human intelligence.
And those limits are about to change forever!
7
💙😇 ⏩️ retweeted
Five people with stage 4 pancreatic cancer now have no cancer showing on their scans.
All 61 patients in this trial came in untreated and got daraxonrasib plus chemo as their first treatment. Tumors shrank in 6 out of 10 of them. Chemo by itself usually holds this cancer back for about 5 or 6 months. With daraxonrasib added, the median was close to 13.
These patients had not been worn down by rounds of chemo yet, so their immune systems were still intact. I think intact immune systems are a big part of why the responses ran so deep.
The drug is great. Delivery is still the problem. Swallowing a pill sends the drug through the whole body. We inject these drugs right into the tumor, so the cancer takes the hit and the immune system stays strong.
Now we need to turn remissions into cures, and the immune system is how we get there.
💙😇 ⏩️ retweeted
HOLY! SCIENTISTS JUST DEVELOPED A NEW CAR-T CELL THERAPY THAT TARGETS TWO DIFFERENT CANCER PROTEINS, SHOWING AN 80% RESPONSE RATE IN PATIENTS WITH ADVANCED BLOOD CANCER! 🤯
The experimental treatment is called TriPRIL, and its early clinical results are incredibly promising.
In a first-in-human Phase 1 clinical trial, 12 out of 15 patients (80%) with relapsed or refractory multiple myeloma responded to treatment.
Even more interesting, 8 patients (53%) had no detectable signs of active cancer in blood, urine and bone marrow assessments.
The technology uses an engineered, trimeric version of a naturally occurring human protein called APRIL (A Proliferation-Inducing Ligand).
APRIL naturally binds to both BCMA and TACI, allowing researchers to build a single CAR receptor capable of recognizing both targets.
This creates another route for engineered immune cells to recognize cancer.
Even if a cancer cell loses BCMA, it may still be detectable through TACI.
And here's another particularly important finding:
9 of the 15 patients had previously received BCMA-targeted CAR-T therapy, but their cancer had returned or progressed.
TriPRIL offers a potentially different approach for patients whose disease has escaped existing treatments.
Researchers also detected engineered TriPRIL cells in some patients' blood more than SIX MONTHS after infusion.
The researchers are planning a Phase 2 clinical trial.
💙😇 ⏩️ retweeted
Congrats to Anthropic & all their amazing scientists! $2T market cap. A new age of medicine begins! 🧬🚀
Riemann Hypothesis is proven
if confirmed and the Riemann Hypothesis is proven, it would be the greatest achievement in the history of mathematics!!!
that's like solving P vs NP, quantum gravity, the origin of life, consciousness, reversing aging, curing all cancers and so much more
which is all coming soon
not even a week after a reusable rocket has been caught falling from the sky
we really are in the singularity and nothing will ever be the same
16
💙😇 ⏩️ retweeted
GOOGLE DEEPMIND RESEARCHERS ARE ALREADY LOOKING BEYOND AGI. AND WHAT COMES NEXT COULD BE EVEN MORE TRANSFORMATIVE. 🤯
In a new essay published by the DeepMind Institutes' researcher Tim Genewein explains 4 potential pathways from AGI to ASI.
The central idea is fascinating:
“AGI might not be the finish line. It could be the beginning of an entirely new era of intelligence.”
Here's what makes this research particularly interesting.
1. AI COMPUTE COULD GROW AT LEAST 10× EVERY YEAR.
Three major forces are compounding:
> Hardware efficiency: approximately 1.5× annually.
> Algorithmic efficiency: training compute requirements have fallen roughly 3–6× annually, while inference efficiency has improved around 3× annually.
> AI hardware investment: approximately 2.5× annual growth.
Together, these factors could increase global effective AI compute by at least 10× annually.
If sustained, that's potentially 10 BILLION TIMES more effective compute over a decade.
2. AI COMPUTATION COULD BECOME EXTREMELY CHEAP.
The essay estimates that continued hardware and software efficiency improvements could reduce the compute cost of a fixed AI workload by approximately 4× every year.
Consider this:
A cognitive workload costing $1 BILLION in AI compute today could cost approximately $1,000 in ten years.🤯
This could fundamentally change the economics of intelligence.
3. BILLIONS OF AI AGENTS COULD WORK SIMULTANEOUSLY.
AI systems can be replicated, deployed across computing infrastructure and coordinated to perform different tasks.
Imagine millions or billions of specialized AI agents working together:
- Discovering new medicines.
- Designing advanced materials.
- Developing better AI algorithms.
- Solving complex mathematical problems.
- Engineering advanced robotics.
- Accelerating scientific research.
The essay suggests that, under its scaling assumptions, the number of equivalent AI agents that could be deployed might eventually exceed the human population.
4. FOUR PATHWAYS TO SUPERINTELLIGENCE.
Tim Genewein identifies four potential routes:
1. Scaling existing models and training data: More compute, better algorithms and increasingly capable models.
2. New technological paradigms: New AI architectures, continual learning, better world models and more efficient reasoning.
3. Massive multi-agent coordination: Thousands, millions or potentially billions of AI agents collaborating as collective intelligence.
4. Recursive self-improvement: AI systems helping develop better algorithms, software, hardware and future generations of AI.
These pathways could reinforce one another, potentially accelerating progress beyond human-level intelligence.
5. RECURSIVE SELF-IMPROVEMENT COULD CHANGE EVERYTHING.
Imagine AI systems conducting AI research, discovering better algorithms and developing more capable successors.
Those improved systems could then accelerate the development of the next generation.
The cycle could repeat.
This is the possibility of recursive self-improvement, although whether it produces runaway progress remains an open research question.
6. RESTRICTING ONE INPUT MIGHT NOT STOP PROGRESS.
The researcher argues that restricting training compute alone won't necessarily stop AI development.
Progress could continue through algorithmic improvements, new architectures, agent coordination and automated AI research.
There are multiple technological pathways, and they could reinforce each other.
DeepMind Institute
Our new essay examines the path from AGI to ASI (Artificial General Super Intelligence)
Read it here: bit.ly/what-comes-after-agi




























