Liliana Xi retweeted
China actually feels behind in e-payments UI now. Apple Pay is more convenient than fiddling with my WeChat to open up a QR code
I think the most likely outcome is that Anthropic neither goes bankrupt nor destroys humanity
It simply stays a valuable company well into the future
But a few years down the line a new startup doing something totally different will come out of nowhere and steal the spotlight
1
54
Liliana Xi retweeted
“Humans are so good at adapting to the new normal that something goes from science fiction to entitlement in approximately 7 days” @Keller
Zipline CEO @keller on how drone deliveries went from "Jesus Christ himself is delivering blood from the sky" to "it's 30 seconds late" in just 7 days:
"You'd go and meet nurses and doctors at a remote hospital and say, now you just send a text message, and an autonomous vehicle's gonna deliver it 10 or 15 minutes later. People are looking at you like you're totally crazy."
"One doctor told me, it's as though Jesus Christ himself is delivering blood from the sky. People would form these huge crowds to watch deliveries for the first few days."
"What is amazing to me is that by day 7, people are 100% bored of it. It's completely normal. I had one doctor come out, and we were watching a delivery together, and she just picked it up, looked at her watch, looked at me and was like, it’s 30 seconds late."
"Humans are so good at adapting to the new normal that something goes from science fiction to entitlement in approximately 7 days."
@zipline
Liliana Xi retweeted
This Thursday I will be giving the opening keynote at RISEx 2026 in Edmonton
Title: The Tashan/Openmind Robot Kindergarten
Abstract: It has always been a dream in reinforcement learning to make robots that learn online by trial and error from their own experience. After all, this is the way people and all other animals learn about their bodies and the world. In robotics, however, learning is almost never online. Instead, the field has focused on learning from data-sets collected from human demonstrations. Learning from experience requires a totally different approach, including anti-fragile robots, skin with tactile sensing, extensive onboard computation, and continual real-time learning algorithms. All these things are difficult and so have rarely been done. They require a substantial commitment of engineering effort and resources, plus determination and patience, plus scientific breakthroughs in reinforcement learning algorithms. The Tashan/Openmind Robot Kindergarten is an ambitious new attempt to do all these things, led by researchers from the Openmind Research Institute in Edmonton and from the Tashan Technology Corporation in Beijing. The work has just begun, and so this talk will focus on the challenge and the ambition. Robots must learn online.
Liliana Xi retweeted
robotics will not explode like ChatGPT.
robotics is a:
1. models problem (YC hivemind overindexes on this)
2. supply chain problem (china FTW)
3. infrastructure problem
this means that the adoption curve will be a lot more gradual.
my favourite robotics companies are that see the bigger picture.
Liliana Xi retweeted
JUST IN: Getty Images files for Chapter 11 bankruptcy after struggling with mounting debt & growing competition from AI-generated images.
Liliana Xi retweeted
I came to realize people liked the french #slopcore without having the lyrics 😅 sorry!
So here is a version with english subtitles!
Liliana Xi retweeted
Le physicien Gabriele Carcassi (@UMich) : je bloquais souvent sur des aspects mathématiques techniques en topologie et l'IA me sort désormais d'affaire en 2 minutes ; il me fallait supplier des mathématiciens pendant des mois, maintenant un seul interlocuteur IA me suffit !
➡️ @AntoineTilloy : "si on est un chercheur raisonnable, il y a forcément une part d'enthousiasme avec l'IA", mais le progrès des modèles est inquiétant car "notre part de guide se réduit" ; "on n'a plus besoin de beaucoup l'orienter, on peut le laisser faire comme un post-doc".
Liliana Xi retweeted
Deep learning is often taught as a sanitized collection of facts and math. But work at the frontier of AI is messy and empirical.
Our new class, Deep Learning Alchemy, aims to teach the art and science of deep learning in the hands-on spirit of Language Modeling From Scratch.
Students build intuition by running experiments on small LMs, then test it through prediction quizzes: given an unseen code diff, can you predict what happens?
deep-learning-alchemy.github…
My co-instructor @kothasuhas and I think there's significant demand for classes that treat students as serious learners who want to be assessed for what they’ve learned, and so we're trying a new exam-heavy approach to the class (7 exams, 85% of the grade, with the remaining 15% in-person discussion assessments).
We’ve designed the class so students can use AI agents to run their experiments and do their assignment, but they then have to demonstrate that mastery on exams without access to the assignment, and any partial credit on the exam has to come through citations to the assignment, incentivizing them to know what’s in their write-up.
The same assignments and quizzes are public for anyone to try (lecture recordings will be released through @StanfordOnline with some delay). We have some very fun assignments and quizzes covering both classic topics (like NQM, run-to-run stochasticity, and scaling laws) and new and timely topics like muP or hyperball. Made by some real experts like @TristanThrush and @wen_kaiyue
Try some quizzes below!
"Adobe is finished" because you vibe coded Photoshop?
As if the Adobe folks hadn't seen it coming!
They had been pivoting to gen AI tools since the early days of the technology
Gen AI will be its moat in the future, not the freaking paint bucket.
They're so far ahead of you
1
187
Why haven't the pragmatic, down-to-earth, "AI is just another technology" realists (Karp, Huang, etc) created a model that can rival the model created by a bunch of death-worshipping, doomsaying, psychosis-suffering AI cultists?
143
The key difference, I think, between open source software and open weight models is that (capable) AI models are simply harder to create, from a technical perspective. Which is why the economy of AI inherently favors centralization more than software in general does
1
105
I've been thinking about how compared to the West, even the Asian notion of last names is individuality erasure. Easier to promote collectivist values if everyone has the same name
45
Liliana Xi retweeted
Replying to @Lisnardinho @BrendanFoody
I still don't think that Malthus has been proven wrong. The Malthusian trap is ultimately an observation about natural selection, and natural selection does not operate on the scale of centuries. Industrialization (or even so-called "singularity") is but a transitional phase


















