xherce (⨍) retweetledi
BREAKING🚨: NASA Mapped the Entire Ocean floor using Gravity from Space.😮
English
xherce (⨍)
10.4K posts
xherce (⨍)
@xherce
Freedom is the possibility of being where you are not expected
London, England Katılım Haziran 2008
1.5K Takip Edilen719 Takipçiler
xherce (⨍) retweetledi
Arran del tancament del Karma, us porto aquestes imatges dels locals nocturns de la plaça Reial, l'any 1994.
A banda del Karma, hi apareixen també el Glaciar, el Pipa i el Jamboree.
Català
AI is becoming weirdly smart, devious, hard to predict, and impossible to measure.
English
Fa dos anys, accedir a un model de llenguatge avançat (LLM) costava 50 $ per milió de paraules processades. Avui: 0,50$. Cent vegades més barat. En vint-i-quatre mesos.
Per visualitzar-ho: si la gasolina hagués caigut al mateix ritme, ompliríeu el dipòsit per... 0,02$.
Català
xherce (⨍) retweetledi
London tech and ai scene heatmap.
Adding more now. Will sync to office spaces next week.
londonmaxxxing.com
English
xherce (⨍) retweetledi
Brain Computer Interfaces are now giving sight back to the blind. A 2mm chip restored sight in 81% of blind patients.
Published in NEJM. FDA reviewing now.
Max Hodak left Neuralink to build this. Here is his story.
garryslist.org/posts/blind-pe…
English
AI will not spread prosperity through technology alone, because institutional bottlenecks, incentives, regulation, and concentration determine who captures gains
English
xherce (⨍) retweetledi
NEW NVIDIA JOB LISTING.
It should tell you all you need to know about where datacenter jobs will go…
IN SPACE.
“What you will be doing:
•Drive architecture for orbital datacenter systems considering everything from the chip out to the satellite and connectivity between satellites”
Link: nvidia.wd5.myworkdayjobs.com/en-US/NVIDIAEx…
English
AI in a nutshell: software deflation vs. hardware inflation, labor displacement, and unresolved debates on timelines. Either you're on the techno-optimist side or in the white-collar pessimist zone. You needn't place yourself in one or the other.
English
xherce (⨍) retweetledi
nanochat now trains GPT-2 capability model in just 2 hours on a single 8XH100 node (down from ~3 hours 1 month ago). Getting a lot closer to ~interactive! A bunch of tuning and features (fp8) went in but the biggest difference was a switch of the dataset from FineWeb-edu to NVIDIA ClimbMix (nice work NVIDIA!). I had tried Olmo, FineWeb, DCLM which all led to regressions, ClimbMix worked really well out of the box (to the point that I am slightly suspicious about about goodharting, though reading the paper it seems ~ok).
In other news, after trying a few approaches for how to set things up, I now have AI Agents iterating on nanochat automatically, so I'll just leave this running for a while, go relax a bit and enjoy the feeling of post-agi :). Visualized here as an example: 110 changes made over the last ~12 hours, bringing the validation loss so far from 0.862415 down to 0.858039 for a d12 model, at no cost to wall clock time. The agent works on a feature branch, tries out ideas, merges them when they work and iterates. Amusingly, over the last ~2 weeks I almost feel like I've iterated more on the "meta-setup" where I optimize and tune the agent flows even more than the nanochat repo directly.
English
xherce (⨍) retweetledi
Just learned about the beautiful personal library of a German mining engineer called Bruno Schröder.
His entire house was covered in custom shelves he built himself, housing his life’s work – a 70,000 book collection.
Bruno died in 2022 at 88, while in the midst of digitally cataloguing his massive collection. Sadly, he had no relatives and the house, including the books, was handed to an estate manager and put up for sale. That’s when his story and these photos surfaced.
English
AI has enormous potential to drive a new wave of economic and societal progress, its success depends far more on institutions, innovation policies, incentives, and historical lessons than on the technology alone.
English
SaaS was built on one beautiful asymmetry: build a thing once, sell it a million times. Fixed cost, variable revenue. That's how you get 80% gross margins and companies worth $300B. The code doesn't cost more to run at user #1,000,000 than at user #1.
AI breaks that. Every inference costs money. Every custom output has a price tag. The "build once, sell forever" model assumes near-zero marginal cost — and that assumption is now wrong. When software gets cooked fresh per user, per task, per moment, the economics look more like a restaurant than a factory. And restaurants don't become trillion-dollar companies.
That's the real threat. Not disruption. Margin compression at the structural level.
English
The CEO of a $95 billion company just said something that should TERRIFY every software executive on the planet.
Patrick Collison, the man who built Stripe, went on TBPN last week and compared the entire software industry to frozen food.
His words: "Software has been created years beforehand, freeze-dried, and then prepared at the moment of consumption."
That era is ending.
His new model for software? Pizza.
Fresh pizza, made to order, right then and there.
Exactly what you need, the moment you need it.
That is the future Collison sees for all software.
What does that actually mean?
It means AI agents will build you custom software in real time.
No subscriptions, bloated dashboards and one size fits all.
Software cooked for you, that moment, then gone.
This is already happening.
Anthropic launched Claude Cowork in January.
Within weeks, $2 trillion in software stocks evaporated.
IBM had its worst trading day in 26 years, legalZoom dropped 20% and the entire SaaS sector is in freefall.
They're calling it the SaaSpocalypse.
The old software model was simple, spend millions building a product, sell it to everyone and collect subscriptions forever.
Fixed cost, infinite monetization and winner takes all.
That game created trillion dollar companies: Salesforce. Adobe, Oracle, Microsoft.
Collison says that game is now breaking.
Why? Because AI introduces real cost at every use.
Inference costs, custom creation costs, every single interaction has a price tag.
No more build once, sell forever and he called it the non-Walrasian software regime.
Translation: The winner take all economics that built Big Software are collapsing.
When every user gets custom software built on demand, there is no single winner.
There are thousands of winners or none.
Think about what this does to pricing.
No more $50/seat/month or enterprise contracts worth millions.
Instead, you pay per task, outcome and for what the AI actually built you.
The entire revenue model of SaaS is being rewritten.
Klarna already ripped out Salesforce and replaced it with AI.
Cursor ditched its paid CMS and built a replacement from scratch.
Companies are doing this now.
The dominoes are falling.
The entire industry is being rewritten in real time.
Milk Road AI@MilkRoadAI
A former Goldman Sachs executive just said something on camera that should terrify every lawyer, doctor, and analyst on the planet. His name is Raoul Pal and he used to move billions on Wall Street. He was asked one question: "How disruptive will AI be?" He said it is the single greatest innovation in human history. Greater than the internet or the electricity. The only thing he compared it to was the splitting of the atom. But here is the part nobody is ready for. He said knowledge is now worth zero. Think about that for a second. Why do lawyers charge $800 an hour? Scarcity of knowledge. Why do consultants bill Fortune 500 companies millions? Scarcity of knowledge. Why did your parents tell you to get a degree? Scarcity of knowledge. AI just destroyed that entire model. A teenager with ChatGPT can now draft legal contracts, build financial models, write code and analyze medical scans. No degree, decade of experience and no six figure student debt. And the numbers already prove it. Employment for workers under 25 in AI-exposed jobs has dropped 13%. Wall Street banks are planning to cut 200,000 jobs. 30% of U.S. companies have already replaced workers with AI. 18 million entry level American jobs could disappear entirely. But here is the part that should keep you up at night. The junior roles where people learn, make mistakes, and develop real judgment are being automated first. Which means the experience that AI cannot replace is the exact experience young workers can no longer get. Pal says humanity now faces a binary choice. Merge with the machines or reject them. There is no middle ground and this is not about robots in factories. This is about the collapse of the entire knowledge economy.
English
xherce (⨍) retweetledi
Defuddle now has a website!
This means you can use Defuddle anywhere to get the main content of a page in Markdown format.
You can simply add "defuddle.md" before any URL, use it via curl, Skills, CLI, or add it to your app via NPM.
x.com/kepano/status/…
kepano@kepano
new project — defuddle > de·fud·dle /diˈfʌdl/ transitive verb > to remove unnecessary elements from a web page, and make it easily readable. Inspired by my work on Obsidian Web Clipper I created Defuddle, an open source library to replace Mozilla's Readability in my projects. Defuddle extracts the main content from web pages. It's a bit more forgiving and tunable to my preferences. Notably, Defuddle uses a website's mobile styles to find unnecessary elements that can be removed from the page. still a work in progress but you can try it out in Obsidian Web Clipper 0.10.9, or install it via NPM
English
xherce (⨍) retweetledi
El músico #JacobCollier es reconocido mundialmente por elevar la improvisación a un nivel de espectáculo comunitario y técnico maravilloso.
En 2025 dirigió a la #NationalSymphonyOrchestra, dándoles notas sección por sección hasta construir una composición completa en vivo.
En la era de la inmediatez y los videos de TikTok, parece que cuesta pararse a ver un video de 15 minutos (lo he editado para eliminar los interludios de aplausos), pero si podéis hacer un leve paréntesis, creo que la experiencia merece la pena.
Siempre me resultará sorprendente que el ser humano tenga la capacidad y sensibilidad de construir y crear cosas tan bellas, y no seamos capaces de dejar de destruir y entrar en conflictos con nosotros mismos.
En fin, como siempre, pido perdón por los posibles errores de traducción y subtitulado... y espero que esto os alegre el día 😉
Español
xherce (⨍) retweetledi
Just when you thought you couldn't possibly like Japan any more 🤩. Can't wait to come to RubyKaigi in October. Such a beautiful, wise country.
Chibi Reviews@ChibiReviews
Japanese PM and the Legendary Manga Author Akamatsu has stated they will not be doing age verification or censorship of the internet. Instead they will teach children the dangers of the internet and still allow freedom of expression. Finally good news! youtu.be/5ZmxhMYTPPg
English
xherce (⨍) retweetledi
Fa 40 anys que Els Joves es va estrenar a TV3!
És una de les sèries més trencadores de la història i li hem dedicat un especial on analitzem com va dinamitar el gènere de la sitcom des de dins i en recordem els millors moments.
El pots escoltar aquí: 3cat.cat/3cat/especial-…
Català