Vinicius Minatogawa

603 posts

Vinicius Minatogawa banner
Vinicius Minatogawa

Vinicius Minatogawa

@vmntgw

🇨🇱 Professor - @pucv_cl; 🇧🇷 Ph.d Mechanical Engineering - @unicampoficial

Chile Katılım Şubat 2009
102 Takip Edilen86 Takipçiler
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
Overleaf had a decade head start and turned it into a paywall. Prism is what Overleaf should've been by now. If you write papers, give it a try this week. The migration takes 5 minutes.
English
0
0
1
24
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
Prism (prism.openai.com): → no compile timeouts → AI integrated natively for editing → free → launched January 2026 Compiled the same paper 4x faster on the first try. No paywall in my face.
English
1
0
0
56
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
Spent 3 hours fighting @overleaf today. Three. Hours. On a paper with 16 references. Then I tried Prism (@OpenAI's new LaTeX editor). Same paper compiled in seconds. A short thread on why I'm done with Overleaf 🧵
English
7
0
2
76
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
@brivael Le business model des labs ne sera pas l'API. Ce sera l'agent qui intermédiera toutes les transactions. Google vend de l'attention humaine. Demain, celui qui contrôle l'agent contrôle la décision d'achat. Ce n'est plus du ads — c'est un péage sur toute l'économie numérique
Français
0
0
0
21
Brivael Le Pogam
Brivael Le Pogam@brivael·
Je viens de réaliser un truc. Amazon et Microsoft jouent aux échecs en 7 dimensions depuis l’arrivée de ChatGPT. Et ils vont avoir tout le monde à la fin. Quand les LLMs ont explosé, les big tech avaient deux options. Construire en interne ou investir dans ceux qui construisent. Google avait les transformers, la data, l’infra, la culture research. C’était à eux de jouer et ils jouent. Microsoft, Amazon, Apple? Aucun des trois n’a la culture pour. Et ils l’ont compris immédiatement. Surtout, ils ont compris un truc fondamental avant tout le monde : il n’y a pas de moat sur les modèles. GPT-4 sort, trois mois plus tard tout le monde est au même niveau. Claude rattrape. Llama rattrape. Mistral rattrape. La course aux modèles est une course de commodité pure. Donc qu’est-ce qu’ils font? Microsoft met $13B dans OpenAI. Amazon met $4B puis $8B dans Anthropic. Tout le monde applaudit les “partenariats stratégiques”. Sauf que c’est pas un partenariat. C’est un put option sur des centaines de milliards de R&D. Parce que le business model des labs ne tient pas. Tu brûles $5-10B par an en compute pour entraîner des modèles qui deviennent commodité en 6 mois. Ton API se fait concurrencer par l’open source. Tes marges se compriment en continu. Il n’y a aucune défensabilité structurelle. Projette toi 3-5 ans. On atteint le plafond cognitif des LLMs. Pas l’AGI. Mais un niveau de cognition artificielle suffisant pour reconfigurer des industries entières. La technologie se stabilise. La différenciation entre modèles tend vers zéro. Et le piège se referme. OpenAI et Anthropic auront brûlé des dizaines de milliards. Financés par le venture capital, le private equity, les retail investors. Le business model implose. Les valorisations s’effondrent. Et au bout du fil avec un preferred equity et un droit de rachat? Microsoft et Amazon. Apple c’est encore plus élégant. Ils ne prennent même pas le risque de l’investissement. Ils attendent que la techno soit mature, ils signent un deal distribution avec Google qui aura fait le travail de consolidation, et ils intègrent tout ça nativement dans l’écosystème. Exactement comme ils l’ont fait avec le search. Zero risque. Zero effort de R&D. Marge maximale. Récapitulons. Tu laisses des startups prendre 100% du risque technologique. Tu laisses les VCs et le marché financer la R&D. Tu récupères la technologie quand elle est mature et que le builder est à genoux. Coût : quelques milliards. Valeur récupérée : des centaines de milliards. Le capital risque pense financer la prochaine révolution. En réalité il subventionne le département R&D des big tech. Et la vraie opportunité de créer des nouvelles big tech via l’IA? Elle ne viendra pas de la couche modèle. Elle viendra dans un second temps, quand la cognition artificielle sera stabilisée et commoditisée, et qu’il sera possible de construire de nouvelles catégories de software par dessus. C’est là qu’il y aura une fenêtre pour bâtir des boîtes gigantesques avec un vrai business model défensible qui tient 20 ans. Mais pour l’instant? On est juste en train de regarder les big tech exécuter le plus beau délégation de risque de l’histoire de la tech. Et franchement c’est absolument génial.
Français
58
134
840
96.8K
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
5/ I’ve enjoyed FIFA for years, but it’s hard to feel confident buying future versions with this uncertainty. Thanks for the good times, EA @EASPORTS — for now, I’ll be going back to eFootball.
English
0
0
0
36
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
4/ After trying many fixes and reading community forums, it seems the issue may be related to EA Anti-Cheat (Javelin). I also found that many other players are facing the same problem.
English
1
0
0
102
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
Thread 🧵 1/ Back in 2012, I decided to leave Winning Eleven / Pro Evolution Soccer (now eFootball) and give FIFA a chance. It wasn’t easy at first, but I adapted — and ended up enjoying FIFA for many years.
English
1
0
0
48
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
@claudeai queria poder entender meus limites de consumo. Ultimamente tenho usado pouco. Hoje fui usar, e 5 interacoes depois me pediu para esperar ate as 22.00. Sou do plano pro faz muito tempo, e antes nao era assim.
Português
0
0
0
3
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
Wouldn’t it be awesome if @manusai played a sound whenever he needs to consult me? I often let him handle things and wander off—and it made me nostalgic thinking about bringing back the classic ICQ notification tone.
English
0
0
1
26
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
É um posicionamento bem ousado vindo do CEO da OpenAI, e mostra como eles estão confiantes nos avanços recentes.
Português
0
0
1
13
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
a segunda parte, onde ele fala de superinteligência - IAs tão avançadas que poderiam revolucionar a ciência e inovação num nível que nem imaginamos hoje. Ele vê isso como caminho para uma era de abundância.
Português
1
0
1
14
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
A postagem do Sam é bem reveladora: ele tá dizendo que finalmente sabem como criar uma IA realmente avançada (AGI). Não só isso - ele prevê que já em 2025 teremos IAs trabalhando junto com humanos e mudando radicalmente como as empresas funcionam. Mas o mais impressionante é
Vinicius Minatogawa tweet media
Português
1
0
1
37
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
@LogTech1999 The Beer Game was always a hit in my classes! Students loved the competitive challenge, and let's be honest - having 'beer' in the name definitely caught their attention. 😄 It made supply chain concepts come alive for them! Heres an online version beergame.masystem.se
Valparaíso, Chile 🇨🇱 English
0
0
0
39
Glenn Koepke
Glenn Koepke@LogTech1999·
At Vector, we have built a team of top technologists fundamentally rethinking how tools in logistics should work. Follow for more latest insights in supply chain technology. withvector.com
English
27
8
144
34.6K
Glenn Koepke
Glenn Koepke@LogTech1999·
In the 1960s, MIT Professor Jay Forrester created something that changed the way we think about supply chains—The Beer Game. This game is a simple simulation, but it messed with the minds of thousands. Here's the game:
Glenn Koepke tweet media
English
112
639
4K
1.2M
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
@paulnovosad That's very true - some books are nearly impossible to translate as something will inevitably be lost. "Grande Sertão: Veredas" by Guimarães Rosa illustrates this perfectly. It is so deeply rooted in its invented language, that even Portuguese speakers struggle with it.
English
0
0
1
111
Paul Novosad
Paul Novosad@paulnovosad·
For any given book, there is some proportion X of the contribution that is lost in translation or updating to modern language. For Ulysses, X is close to 100%. For political philosophy, it is much much lower. X also depends on what you want to get out of it.
English
3
1
8
3.6K
Vinicius Minatogawa
Vinicius Minatogawa@vmntgw·
@emollick My teachers used to say before evaluations 'memorize things, you won't always have books with you'. Now we kind of have. We should focus on understanding core concepts deeply enough to collaborate with AI and evaluate its outputs. We have to change learning b4 change evaluations
English
0
0
1
362
Ethan Mollick
Ethan Mollick@emollick·
Reminder for the new semester When researchers secretly added AI-created papers to the exam pool: “We found that 94% of our AI submissions were undetected. The grades awarded to our AI submissions were on average half a grade boundary higher than that achieved by real students.”
Ethan Mollick tweet mediaEthan Mollick tweet media
English
21
122
527
65.2K