Colin Lewis (@ColinWPLewis) header
Colin Lewis (@ColinWPLewis) profile picture

#AI & Data Science. AI to augment economic decisions. Author of The AI Leadership Code. #Triathlon. Relentless iteration. 科林. Deep interest in China Innovation

Joined September 2013
Photo by Colin Lewis (@ColinWPLewis)Photo by Colin Lewis (@ColinWPLewis)Photo by Colin Lewis (@ColinWPLewis)Photo by Colin Lewis (@ColinWPLewis)Photo by Colin Lewis (@ColinWPLewis)Photo by Colin Lewis (@ColinWPLewis)Photo by Colin Lewis (@ColinWPLewis)Photo by Colin Lewis (@ColinWPLewis)Photo by Colin Lewis (@ColinWPLewis)Photo by Colin Lewis (@ColinWPLewis)
Selfie moment with @sama - Incredibly humble and smart. Privileged to listen to Sam talk today.
2
1
18
3,722
Yes, I did, because I understand how valuable @JensenHuang is to our country.
JUST IN: Former White House Press Secretary Katie Zacharia demands Nvidia reinstate Jensen Huang as CEO "immediately."
2,914
1,768
215
24,728
1,102,919
Colin Lewis retweeted
卧槽,TMD 全世界都在踩黄仁勋的时候 亿万富翁 Bill Ackman 站出来替他说话了,理由让人瞬间破防 Ackman 在采访里说,他女儿一直有非常严重的健康问题,和脑部损伤有关,他私下找黄仁勋求助,想把英伟达的团队和自己一个专门帮助脑损伤病人的项目连起来,老黄二话没说就帮了, Ackman 的原话是,他对这件事非常 wonderful 主持人问他为什么偏偏这个时候出来表态,他说自己信奉无罪推定,媒体在每一个案子里,都是事实还没搞清楚就先把人定了罪 据 The Information 报道,董事会起初并不想调查,最后让老黄下台的是他和第二位直属下属的关系,一边是媒体一篇接一篇往外爆,一边是亿万富翁公开站队,同一个人在两拨人眼里完全是两副面孔 打工人都懂这个感觉,你平时在公司帮过的人不少,可等你出事那天,有几个敢站出来替你说一句话,大部分人只会默默把你从群里屏蔽,就像下暴雨时你借过伞的人很多,等你自己淋湿了,回头给你递伞的恐怕只有一两个
252
1,921
75
14,374
1,141,228
Colin Lewis retweeted
Walking through the Independence March in Warsaw, the sheer normality of it hit me. Poles of all ages, families included, openly celebrating their sovereignty and waving their flags with pride. No one was looking over their shoulder. No one was worrying about showing their face and losing their job or being arrested. They found it hard to believe that back in the UK the same instincts get you labelled far-right and can land you in a police station, for a tweet, for flying the Union Jack on the steps of a council building, or simply for saying you want control of your own borders. FULL VIDEO COMING SOON To support my work, you can donate, or buy Issue 1 and 2 of my free speech magazine Proper-Gander on my website: espeaksfreely.com/category/a…
40
355
6
1,710
18,346
.".. plausibly as early as next year), we’ll have AIs that are at least as good as the best human experts in every field." agreed @dwarkesh_sp dwarkesh.com/p/rsi
93
just finished Chemotherapy. I would be happy if you wish me Health and Long Life 🙏🙏
1,578
717
85
15,179
246,191
Najlepsze podsumowanie 🎯 Francuski dziennikarz i publicysta, Alexandre Mendel, skomentował Marsz Niepodległości w Warszawie👇 "Bilans marszu na ulicach Warszawy: 500 000 osób, żaden sklep nie został splądrowany, żaden policjant nie został pobity, żaden samochód nie został spalony. Dziwne dla Francuza" - podsumował. Dlatego bądźcie jak Polska 🇵🇱 Marsz Niepodległości to największa i najwspanialsza manifestacja patriotyczna w Europie!
266
2,712
71
10,604
96,154
Before leaving Poland, I paid a visit to the site of the tragic incident in Krakow. I expressed my deepest condolences and greeted police officers guarding the site to extend the solidarity from the Japanese people. Poland, we are with you. @PLinTokyo @AmbasadorJP
386
1,710
28
26,403
376,067
Colin Lewis retweeted
Dostoevsky exposes you. Tolstoy widens you. Kafka unsettles you. Woolf slows you down. Camus frees you. Orwell warns you. Austen reads you better than you read yourself. Hemingway strips you bare. Nietzsche dares you. Márquez enchants you. You don't read these people. YOU SURVIVE THEM.
99
604
169
3,387
317,855
Colin Lewis retweeted
This chart!
7
23
17
102
64,379
Prezes znanej polskiej firmy i paru polskich naukowców z czołowej uczelni cytuje artykuł z jakiegoś anonimowego konta bez źródła i wyciąga z niego daleko idące wnioski pomimo, że tekst jest z 2024 rok, a modele są ewidentnie coraz lepsze. Słabe to. teapawt.org/CrazyShyyt/status/2118…
ChatGPT has now a big problem. Researchers at Oxford exposed a massive threat to large language models.” It’s called "Model Collapse." And it explains why the internet is about to become a toxic waste dump for Large Language Models. Right now, AI companies are scraping the entire web to train their next-generation models. But there’s a massive problem. The internet is no longer just human data. It is rapidly filling up with AI-generated text, images, and code. What happens when an AI trains on data generated by another AI? A new paper in Nature just gave us the answer. And it is terrifying. When AI models are recursively trained on AI-generated data, they suffer from irreversible brain damage. The researchers found that the model starts forgetting the "tails" of the original data distribution. The rare, weird, unique, and complex human nuances completely disappear. It's like making a photocopy of a photocopy. Every generation loses resolution. The output gets flatter, blander, and more generic, until the model completely collapses in on itself. For years, we thought the biggest bottleneck for AI was compute, getting enough H100 GPUs from Nvidia. This paper proves the real bottleneck is us. If we keep flooding the internet with synthetic AI slop, we will poison the very data pool required to build smarter models. The ultimate irony? We thought AI was going to make human creators obsolete. Instead, this research proves that verified, genuine human data is about to become the most valuable commodity in the tech industry.
7
5
89
4,523
I turned @shaunmmaguire's brilliant bottom-up model into an interactive site, so that anyone can play around with the underlying assumptions. Few-shotted by combining @bot (for research/structure) with @ChatGPT (for the build itself) See the site, Orbital Thesis -> orbital-thesis-spacex-lab.fl…
Over the last week I built a bottoms up SpaceX financial model through the end of 2028 The numbers that came out are insane There are wide error bars given how quickly things are changing but I wanted to distill something complicated into something that can at least be debated * This is not investment advice!!! ** Sequoia is a major SpaceX shareholder *** Past performance is not indicative of future results
1
7
1,823
Colin Lewis retweeted
Not everything made with AI is slop. Not everyone using AI lacks creativity. Some people are creating beautiful things, telling incredible stories and bringing ideas to life. Your inability to appreciate it doesn't make those creations worthless.
Englishman in New York
81
149
13
642
36,742
I am releasing my full documentary on the riots in Paris. I spoke to the rioters themselves and you'll be able to see how they rationalize their crimes. In this video you'll hear from rioters that they want to smash everything because the gas prices are too high. You'll hear the rioters say that they are going to "f*ck Marine Le Pen, that wh*re." You'll hear from others that they just want to loot the big businesses. Either way, it was nothing but people encouraging their countrymen to destroy what others have created so they can be "seen." This was continuously covered and is still is covered as a student protest. It's a lie through omission, yes it may have started that way but the people I spoke to were not interested in better textbooks, they were interested in burning down Paris until they get what they want which they say is a lower cost of living. My message back to America is this: you would be naive to think this is not what is coming next for us. Yes we had riots in 2020 for a very different reason, no individual action spurred the riots I saw here, this is just the rise of communism, islam, and decades of unvetted mass migration that is now turning violent. Paris is one of the most beautiful cities I have ever been to and it's heart wrenching to see the tidal wave that has finally hit it. The best way to support my reporting is by subscribing on X.
178
2,848
105
8,501
144,985
Colin Lewis retweeted
Tyler Cowen: 'I am not even sure what to say at this point. We need a radical revaluation of this work, and of the associated policy arguments. More generally, this is only a single step in the story of how AI is going to revise economic science, and indeed past research more generally. As I stress in many of my recent talks, there will be significant gains and losses in status, with the losses often outweighing the gains. I see this as one of those cases.'
98
215
75
1,977
311,762
Colin Lewis retweeted
I know it’s been said so many times, but I just love @Waymo! I have to admit when Larry and Sergey would obsess over self-driving cars, I was skeptical, but it’s just so great.
16
23
3
404
43,248
Colin Lewis retweeted
Absolutely stunning essay by @DouglasKMurray about the crisis in France. It includes this quote. thefp.com/p/douglas-murray-f…
130
1,348
83
7,319
287,929
I have been a paid subscriber to ChatGPT from day 1... the last two months I have not used it at all. Time to ditch the monthly subscription!
112
Colin Lewis retweeted
Replying to @KonradKrajewsk6
Cześć, bardzo dziękuję za Twój wpis. Staram się wszystko wyłapać i choć kilka słów napisać. Motywacja tego modelu była prosta: stworzyć użyteczne narzędzie, otwarte. Na codzień tworzę wiele innych modeli komercyjnie i tam otrzymuję pieniądze. Przyszedł taki czas gdzie stwierdziłem, że akurat ten model oddany open będzie świetnym pomysłem na promowanie "możemy wiele" - mówiłem o tym w wywiadach (zawsze twierdziłem, że może GPT nie zbudujemy ale z optymalizacji zawsze byliśmy bardzo dobrzy). Co do Gemma. Tak. Flagowy model jest na Gemma. Jeśli ktoś myśli, że wystarczy ot tak finetunować to zapraszam do wytrenowania swojego modelu i poddania go ewaluacji na JevBench. Jeśli zrobi TOP#1 to z mojej strony pójdą oklaski. Szczegóły na temat projektu dla tych co mówią to tylko LoRA na Gemma albo tylko finetuning na Gemma umieściłem w ostatnim filmie. To ładnych kilka lat doświadczeń w R&D AI oraz wylany pot i łzy przy trenowaniu tysięcy modeli (zarówno wizyjnych jak i LLM'ów). Czy można poprawić model? Tak. Jutro wydaję zoptymalizowany inference engine (bardzo duże przyspieszenie). Backlog dla wersji basal-omni 2.0 (również na Gemma) to kilkadziesiąt pomysłów do zrealizowania. Na Bieliku już nie będzie kolejnej wersji (tylko fixy po wersjo omni-2.0). Co dalej? Teraz całkowicie jestem skoncentrowany na rozwoju basal'a (jako open) oraz pracy zawodowej (AI w biologii, również wsparcie w zakresie systemów satelitarnych SAR). Dziękuję za wpis. Bardzo doceniam.
3
7
141
1,731
Colin Lewis retweeted
An Anthropic researcher just told a room full of PhD students that their career plan no longer exists His 3 pieces of advice for students: 1. Use AI every day. If you're on the exponential, you'll never fall behind it. 2. Choose projects that get better when compute scales. The rest will be solved before you finish them. 3. Stop planning your career as if it were stable. The safe option no longer buys safety; it no longer exists. And a phrase that sums it all up: "Writing is thinking. If you outsource your writing, you outsource your thinking." The talk is titled "The End of the Anthropocene"
38
196
31
1,635
234,455