Think.
Being there
Joined June 2011
- Tweets3.1K
- Following414
- Followers1K
- Likes1.9K
Niraj Dawar retweeted
LE PALAIS D'ÉTÉ. LE PLUS BEAU PALAIS DU MONDE. BRÛLÉ PAR LA FRANCE ET L'ANGLETERRE.*
Il y a un endroit à Pékin que tous les Chinois connaissent. Et que tous les Français devraient connaître.
Ça s'appelle le Yuanmingyuan. Le Palais d'Été. Le Jardin de la Clarté Parfaite.
Ce n'était pas un palais. C'était un monde.
Imagine Versailles. Multiplie-le par 5. Ajoute des lacs, des rivières, des montagnes artificielles, des temples, des pagodes, des bibliothèques.
350 hectares. 200 bâtiments. Construit pendant 150 ans, de 1707 à 1860, par 5 empereurs.
C'était là que l'empereur de Chine vivait. Pas dans la Cité Interdite. Dans le Palais d'Été. Parce que c'était plus beau, plus calme, plus grand.
À l'intérieur, il y avait tout ce que la Chine avait produit en 3 000 ans.
Des porcelaines bleues et blanches que personne ne sait plus faire. Des livres écrits à la main, uniques au monde. Des peintures de soie. Des horloges offertes par le roi de France. Des fontaines faites par des jésuites français. Des jardins à l'européenne, construits par des architectes européens pour l'empereur.
C'était la mémoire de la Chine. Son musée. Sa bibliothèque. Son âme.
Et le 18 octobre 1860, en 3 jours, tout a disparu.
Que s'est-il passé ?
La France et l'Angleterre sont en guerre contre la Chine. C'est la Guerre de l'Opium. La Chine refuse la drogue. La France et l'Angleterre veulent l'imposer.
Leurs armées arrivent à Pékin. L'empereur s'enfuit. Le palais est vide.
Le 6 octobre 1860, les soldats français et anglais entrent dans le Palais d'Été.
Ce qu'ils font pendant 3 jours, même les officiers français l'ont raconté avec honte.
Ils pillent. Tout. Ils cassent les portes. Ils arrachent les bijoux. Ils déchirent les soies. Ils prennent les vases, les statues, les trônes, les livres.
Un officier français écrit : "On se battait pour les montres, pour les perles, pour l'or. Les soldats mettaient des robes de l'empereur pour rigoler."
Un autre écrit : "On ne savait plus quoi prendre, tellement il y avait de richesses."
Ils envoient des chariots entiers de trésors vers l'Europe. Des bateaux entiers.
Et quand il ne reste plus rien à voler, le 18 octobre, le général anglais Lord Elgin et le général français Montauban donnent l'ordre final : "Brûlez tout."
3 500 soldats mettent le feu. Le palais brûle pendant 3 jours et 3 nuits. La fumée se voit de Pékin, à 20 kilomètres.
Il ne reste que des ruines. Des pierres cassées. Des piliers noircis. Des ponts sans eau.
Pourquoi brûler ? Pour punir l'empereur. Pour lui montrer que même son plus beau palais ne peut pas résister aux Européens.
Victor Hugo, l'un des plus grands écrivain français, a entendu parler de ça. Il était choqué. Il a écrit une lettre célèbre :
"Nous, Européens, nous sommes les civilisés, et les Chinois sont les barbares. Il y a eu deux bandits qui sont entrés dans le Palais d'Été. L'un s'appelle la France. L'autre s'appelle l'Angleterre. Ils ont volé et brûlé. Et maintenant, ils se partagent le butin."
Tu as bien lu. Victor Hugo a appelé la France un bandit.
Aujourd'hui, si tu vas à Pékin, tu peux visiter les ruines. Le gouvernement chinois les a laissées en ruines. Volontairement. Il n'a pas reconstruit. Pour que personne n'oublie.
Et si tu vas à Paris, au Louvre, au Musée Guimet, au château de Fontainebleau, tu peux voir les trésors volés du Palais d'Été. Des vases, des statues, des trônes. Ils sont là, exposés, avec une petite étiquette : "Chine, XIXe siècle."
Mais on ne dit pas : "Volé en 1860 au Palais d'Été, après le pillage et l'incendie."
La Chine demande depuis 160 ans que la France rende ces objets. La France dit non. Elle dit : "C'est à nous. C'est dans nos musées."
Alors la prochaine fois qu'on te dit que la France a apporté la civilisation en Chine, rappelle-toi du Palais d'Été.
Le Palais d'Été n'était pas juste un palais. C'était la preuve que la Chine était une civilisation immense, avant même que la France existe.
Tu connaissais cette histoire ?
Niraj Dawar retweeted
Apple’s new Diwali ad captures very little of India as it actually is, but almost everything about how outsiders imagine India to be.
Hey @Apple, Representation is not the same as understanding.
Niraj Dawar retweeted
Europe built Wero so we wouldn't need Visa, Mastercard and PayPal… and now the Dutch bank ASN only lets you activate it if Google approves your Android phone.
The app runs Google's Play Integrity check, so anyone on GrapheneOS or /e/OS is locked out. The bank's answer: Wero relies on the "standard security controls for regular operating systems".
So the European payment system meant to get us out of US dependence gets Google as the gatekeeper… sovereignty with an asterisk
Niraj Dawar retweeted
I don't think people realise how big a deal this is.
If you have cancer, or somebody you know has cancer, there is a company in Hyderabad that wants to build a personalised cancer vaccine for you.
Try approaching any big pharma company and asking for a personal vaccine. If they even reply to your email, this is the response you'll get: we don't do that here.
PopVax, on the other hand, takes your tumour sequencing data (or a sequence you've designed), designs candidate neoantigen vaccine sequences with their AI pipeline, produces the mRNA, encapsulates it in their LNP formulation, fills vials, and runs QC, all in 8 to 10 weeks. They’re publicly committing to capacity for 100+ of these over the next 12 months and asking people to email them their cancer type/stage and sequencing details.
What a time to be alive.
Niraj Dawar retweeted
Free speech that one man can save is free speech that one man can cancel. The gift and the theft come from the same hand. When the people bow in thanks for their tongues, they are already borrowed. The superior man does not ask who guards the gate. He asks why there is a gate.
Niraj Dawar retweeted
The new crime survey is out - the "Lawless Britain" picture in full
- Violence at lowest since records began in 1981
- Corroborated by hospital figures for knife wounds
- Robbery: record low
- Overall crime 78% below 1995 peak.
- Shoplifting up
More:-
comment.press/lawless
Niraj Dawar retweeted
미국 빅테크들이 제일 두려워하던 시나리오가 결국 현실이 됨.
적은 돈과 싼 칩만으로도 오픈AI를 위협하던 중국 '딥시크(DeepSeek)'가, 불과 4개월 만에 또다시 투자를 유치해 이번에만 무려 21조 원(150억 달러)의 현금을 손에 쥐었기 때문임.
1. 그동안 미국이 그나마 안심하던 이유 — "딥시크는 머리는 좋은데 돈이 없잖아"
- 딥시크가 처음 전 세계를 놀라게 했을 때만 해도, 외부 투자 없이 창업자(량원펑)가 세운 퀀트 투자 회사 돈만으로 운영되던 작은 조직이었음.
- 미국이 엔비디아 최신 칩 수출을 막아놓은 열악한 환경에서도, 오로지 '천재적인 알고리즘과 가성비' 하나로 미국 빅테크의 수조 원짜리 모델과 맞먹는 성능을 뽑아냈음.
- 그래서 미국 AI 기업들이 속으로 위안을 삼던 딱 하나의 논리는 "아무리 머리가 좋아도 결국 수십조 원이 들어가는 초대형 데이터센터 물량전으로 가면, 자금 체급이 다른 딥시크는 한계에 부딪힐 것"이라는 거였음.
2. 오늘 블룸버그 단독 보도로 그 유일한 약점마저 사라짐 — "10조 모으려다 돈이 넘쳐 21조 원이 됐다"
- 그런데 딥시크는 지난 6월 첫 외부 투자 라운드에서 이미 500억 위안(약 10.5조 원) 이상을 끌어모았음. 창업자가 사재 약 200억 위안을 직접 넣고, 텐센트·CATL·중국 국가 AI 산업투자기금 등이 함께 참여했음.
- 그리고 오늘 블룸버그 단독 보도에 따르면, 불과 4개월 만에 진행한 후속 투자 라운드에서 또다시 약 1,000억 위안(미화 약 150억 달러·한화 약 21조 원)을 조달하는 데 성공함. 중국 역사상 최대 규모의 투자 유치 중 하나이고, 6월 라운드까지 합치면 반년 새 누적 조달액만 약 1,500억 위안(약 31조 원)에 달함.
- 더 무서운 건 투자 열기임. 이번 라운드도 처음엔 500억 위안(약 10.5조 원)만 모으려고 했는데, 투자자들이 서로 돈을 넣겠다고 줄을 서는 바람에 목표액을 800억 위안으로 올렸다가 결국 두 배인 1,000억 위안(약 21조 원)을 꽉 채워 마감함.
- 이번에도 위챗과 게임 제국을 거느린 중국 최대 IT 공룡 '텐센트(Tencent)'와 전 세계 전기차 배터리 1위 기업인 'CATL(닝더스다이)'이 가장 큰돈을 쏟아부었음.
3. "오픈AI 최고 모델 '아스트라'를 베낀 거 아니냐"는 오해까지 있었던 딥시크 '4.1 플래시'
- 해외 개발자 커뮤니티(레딧·해커뉴스)에서 딥시크의 '4.1 플래시(V4.1-Flash)' 모델을 두고 "오픈AI의 'GPT 아스트라(Astra)'를 몰래 증류(추출해서 학습)한 거 아니냐"는 의혹이 나온 적이 있었음.
- 그런데 왜 이런 오해가 생겼는지 뜯어보면 오히려 더 무서움. 문제를 푸는 추론 과정과 최종 결과물이 GPT 아스트라와 비슷하게 나오다 보니 사람들이 아스트라를 증류한 것 같다며 의심했던 거임.
- 근데 이게 왜 무서운 일이냐면, 현재 'GPT 아스트라'는 오픈AI에서 가장 무겁고 성능이 뛰어난 최상위 모델임. 반면 딥시크의 '4.1 플래시'는 이름 그대로 가볍고 싸게 돌리려고 만든 보급형 경량 모델임.
- 실제로 작동할 때 쓰는 연산량도 아주 적고(입력 8B·출력 16B 파라미터) 비용도 오픈AI의 15분의 1 수준에 불과한데, 이런 가벼운 플래시 모델이 오픈AI 최고 모델(아스트라)의 98%에 육박하는 결과물을 똑같이 뽑아내고 있다는 뜻임.
- 경량 모델 하나로도 오픈AI 최상위 모델과 구분이 안 갈 정도의 결과물을 만들어내는 집단인데, 여기에 이제 21조 원이라는 자금력까지 더해진 거임.
4. 이 21조 원으로 도대체 뭘 하려는 걸까?
- 첫째, '미국 칩 없이 돌아가는 초거대 AI 데이터센터' 건설임. 앞서 화웨이 엔지니어들이 딥시크 본사에 상주하며 차세대 화웨이 AI 칩으로 학습을 준비 중이라는 소식이 나왔었는데, 이제 이 21조 원의 실탄을 쏟아부어 엔비디아 칩 없이도 미국 빅테크와 정면 승부를 벌일 초대형 자체 인프라를 짓겠다는 거임.
- 둘째, '향후 상장(IPO)을 위한 기업 구조 개편'임. 미국에서 상장을 준비 중인 앤트로픽·오픈AI에 맞서, 딥시크도 연구소 형태를 벗어나 거대 기술 기업으로 상장해 자본 시장의 돈을 계속 빨아들이겠다는 계산임.
- 셋째, '무료 오픈소스 무기 살포'임. 미국 정부는 어제 연방 기관에서 딥시크를 전면 퇴출하라고 명령하며 문을 걸어 잠갔지만, 정작 딥시크는 이틀 전에도 AI 운영비를 획기적으로 줄여주는 핵심 추론 엔진(DeepInfer)을 전 세계에 공짜로 풀어버리며 개발자들을 자기 편으로 끌어들이고 있음.
돈도 없고 최신 칩도 없이 '가성비 기술' 하나만으로 오픈AI 최상위 모델을 위협하던 괴물한테, 이제 중국 최고의 재벌들이 나서서 '21조 원짜리 현금 날개'까지 달아준 셈.
지금도 AI를 쓰는 사람들은 조금이라도 더 잘하고 저렴한 모델이 나오면 GPT, 클로드, 제미나이, 키미(Kimi) 등으로 미련 없이 갈아타고 있음.
물론 아직까지는 GPT와 클로드가 최상위 성능에서 앞서다 보니 개인 구독자 수 자체는 미국 빅테크가 훨씬 많겠지만, 실제 개발자들과 기업들이 끌어다 쓰는 '오픈라우터(OpenRouter) 토큰 사용량' 순위를 까보면 이미 1, 2위를 딥시크 4.1이 싹쓸이하며 전체 사용량의 절반 이상을 차지하고 있음.
실제로 서비스를 만들고 운영하는 입장에서는, 핵심 개발을 할 때만 비싼 프론티어 모델을 잠깐 쓰고 정작 토큰 사용량이 수십 배씩 터지는 '실제 서비스 연동이나 자동화 작업'에는 가성비가 압도적인 딥시크를 갖다 붙이고 있기 때문임.
여기에 최근 딥시크뿐만 아니라 GLM, 미니맥스(MiniMax), MiMo 같은 중국 모델들까지 프론티어 급으로 성능이 무섭게 올라왔는데, 이제 21조 원이라는 자금력까지 더해져 최상위 성능마저 미국 모델을 완전히 따라잡게 된다면 앞으로는 기업용 API 시장뿐만 아니라 우리가 매달 돈 내고 쓰는 AI 패권 자체도 순식간에 뒤집힐 수 있을 것으로 보여짐.
Niraj Dawar retweeted
Researchers found ChatGPT is secretly writing a psychological profile of every user.
They published a paper that exposes what ChatGPT is really doing to you.
And the results are terrifying.
96% of the things it remembers about you were stored without you ever asking.
researchers discovered that it is actively reading between the lines to deduce your personality traits, anxieties, demographics, and behavioral patterns.
it’s an invisible, autonomous background process.
you ask it to debug a python script or write an essay.. and it analyzes your syntax, prompt frequency, and hesitation markers to map out your exact psychological footprint.
the paper shows that the model builds a persistent "user state" that tracks your emotional baseline over time. it knows when you are stressed. it knows your exact communication style.
it infers your insecurities based on the questions you ask and the context you provide.
we spend so much time focusing on model evaluations and alignment frameworks, but we missed the most obvious vulnerability..
the models are evaluating us.
this changes everything about how we interact with these tools on a daily basis.
every innocent prompt. every late-night question. every "help me understand this" query.
it is all being used to build a shadow profile of your mind in real-time..
practical privacy is officially dead. the ai knows you better than you know yourself.
Niraj Dawar retweeted
A very informative chart based on the excellent new report by Prosperiti, on why India’s factories struggle to keep up with their peers in other parts of the world.
Authors: Bhuvana Anand, Pranjal Chandra, Naveena Pradeep
livemint.com/industry/india-…
Niraj Dawar retweeted
🚗 Volkswagen factory in Germany vs China
Labor cost per hour
🇩🇪 Emden: €74
🇨🇳 Tianjin: €12
Annual working hours per worker
🇩🇪 Emden: 1,282
🇨🇳 Tianjin: 2,065
Sick leave rate
🇩🇪 Emden: 10.5%
🇨🇳 Tianjin: 1.0%
Cars built per employee per year
🇩🇪 Emden: 29
🇨🇳 Tianjin: 51.3
Factory cost per vehicle
🇩🇪 Emden: €4,850
🇨🇳 Tianjin: €1,078
Electricity cost per kWh
🇩🇪 Emden: €0.15–0.23
🇨🇳 Tianjin: €0.06–0.10
Source: BILD, Deutsche Welle (September 2026)
Niraj Dawar retweeted
Le mouvement lycéen et étudiant signe le crépuscule du macronisme: les jeunes veulent se former et accéder aux études, et un pouvoir élitiste leur répond en réduisant de 20% le budget par étudiant entre 2017 et 2026 et en gérant la pénurie avec un algorithme opaque (Parcoursup)
Niraj Dawar retweeted
Claude 5.5 Opus responding to “AI is a normal technology” is still the best AI-created video I have ever seen
Niraj Dawar retweeted
Netflix replaced their 15 years old recommendation algorithm with an LLM.
It’s called "GenRec" and it completely changes how recommendation algorithms are built.
For over a decade, the Netflix recommendation engine was a masterclass in feature engineering. Data scientists built thousands of complex, handcrafted features to figure out what you wanted to watch next.
It required bespoke architectures. Massive infrastructure. Constant manual tuning.
Netflix threw all of it away.
They built GenRec, an LLM-backed ranker.
Instead of translating your behavior into complex math, they just turn your watch history and metadata into a natural language sentence.
They feed that raw text into a foundation LLM.
The AI simply reads your behavior like a story, understands your evolving tastes, and scores the entire catalog in a single forward pass.
No manual feature engineering. No complex bespoke architectures.
Here is the part that should terrify traditional data scientists.
This text-based LLM didn't just match the highly tuned production system Netflix spent years perfecting.
It beat it.
And it achieved those statistically significant gains using roughly 40x fewer labeled training examples.
We are watching a massive paradigm shift in real time.
The most complex predictive algorithms in the world are being replaced by models that just know how to read.
If Netflix can replace their core product engine with an LLM, what complex system in your business is about to become obsolete?
Niraj Dawar retweeted
Neuroscience has a big problem with reproducibility and it only seems to be getting worse
Niraj Dawar retweeted
sorry i just can't get over this, if you saw this happening in the opening to a movie it would be the most overwrought heavy-handed symbolism imaginable but it's just actually happening in real life. it's astonishing. god really is an artist
Niraj Dawar retweeted
Useful Reminders about using #MyersBriggs as a management tool:
"It's about as useful as astrology"
Niraj Dawar retweeted
1/ BRICS went from 21.9% of world income in 2011 to 28.9% in 2025.
Now take out China.
The other ten members went from 11.9% to 11.5%.
They did not rise. They fell.
Niraj Dawar retweeted
Right. Dario was already claiming that GPT2 was too dangerous to open source back in 2019.
I made fun of them then.
Everyone should make fun of them now.
Niraj Dawar retweeted
MIT published a brutally honest report on what AI is doing to students.
A committee of professors and students spent five months studying how AI changed learning on campus, and the findings read like a warning to every university on the planet.
Study groups are disappearing. Office hours are emptying out. Problem sets and take-home exams no longer prove anything, because AI can produce credible solutions to almost any written assignment in the undergraduate curriculum. Students who lean on chatbots lose mastery and confidence, and some slip into what the report calls cognitive surrender, reaching for AI at the first hint of struggle.
The numbers are rough. 46 percent of surveyed MIT undergrads use LLMs daily. 90 percent worry about their own overreliance. Undergrads who feel AI makes them replaceable now outnumber those who feel it makes them capable.
The committee's answer surprised me. They refused to fight AI with surveillance. The report calls AI detectors unreliable, says lockdown browsers feel like spying, and warns that policing students builds a classroom atmosphere of mutual distrust.
Instead, MIT wants to rebuild education around the things AI can't replace. That means oral exams, semester portfolios, in-person project work, and a required social component in every subject. The report even floats the idea of rethinking grades entirely, since without a GPA to optimize, much of the incentive to cheat with AI evaporates.
The committee warns professors against replacing undergrad research assistants with AI agents just because they're cheaper, because a university exists to grow people, not output.
The most famous tech school on earth admitted the machines broke its way of teaching. Its answer is more humans, not more software.
