O vídeo do MC Cabelinho no Revelando no Instagram tem 31.5 milhões de views em 24 horas (mais uns 20 milhões se contarmos cortes e teaser). O do Porchat sobre o laranjal no mesmo canal teve 14.4 milhões (mais 25 milhões contando cortes/teaser) O do Wagner Moura sobre Terras Raras teve 50.5 milhões, 55.2 milhões em repostagem, mais 34 milhões em cortes. Revelando é uma página que tem um décimo dos seguidores do Nikolas, e foi criada há 7 semanas. Os números são suspeitos? Pra mim nem um pouco. É um conteúdo bem produzido, único na esquerda, que — aqui o mais importante — é compartilhado por enorme parte dos atores políticos relevantes. E por isso o número de seguidores é menos relevante. O Brasil é de longe o país com mais engajamento online do mundo. Celebridades de reality em outros países não tem nem remotamente o nível de engajamento que temos aqui com BBBs e seus admins. Quando o X caiu no Brasil em 2024, uma enorme parte das "stan accounts" (fã-clubes de artistas gringos") caiu junto. No Brasil, há muitos "políticos" que na verdade são produtores de conteúdo, e que por acaso batem cartão no Congresso. Os que fazem conteúdo único, distribuído por muita gente relevante, tem alto engajamento. Em outubro, é gente falando sobre o assunto mais engajante para a plateia mais engajada do mundo. Toda análise e "estudo" precisa levar essas coisas em conta, e fazer comaprações dentro do mesmo universo comparável.

Worth repeating: Expect an avalanche of new software because the dam has broken, one person who knows what they want, and understands UI design can make a big complex app in 2-3 months. Over a weekend you can snap together an app from common UI concepts. Think of making software the same way you play with LEGOs.

Pros colegas que até hoje passam pano pra fake da Aparecida, vamos fazer um exercício: Imagina se algum jornalista da Globonews fala que ouviu de cristãos que "estão com medo de que Lula irá fechar as suas igrejas". E fala isso sem nada além. Sem pontuar que é "uma proposta completamente absurda, fora da realidade". Concordamos que isso seria má prática, né? Que seria utilizar a grife da emissora pra lavar uma fake. Então. Os bolsonaristas agora estão com essa campanha de "Lula não feche minha igreja", que é uma histeria fake. Ela começou depois que acharam tuítes da "nova cara da comunicação do PT" estupidamente "crente-fóbicos". Pedindo inclusive pra fechar igreja, entre outras coisas absurdas típicas de pessoas "jovens e politizadas" que vivem em uma bolha. Ninguém são acha que isso pode acontecer no governo Lula. Talvez extremistas de Direita achem. Mas simplesmente passar pra frente isso sem averiguar a plausibilidade da ameaça é, novamente, ser inconsequente. Da mesma forma que a esquerda defendeu a jornalista da Globonews, bolsonaristas iriam dizer que tá certo em noticiar. Afinal o PT, ao incensar a jovem e definí-la como exemplo, concordam com o que ela fala, então há um fundo de verdade aí. É a clássica culpa por associação, como fizeram com o cara do projeto de lei de 2007. Os dois lados deveriam parar, mas não vão. Quem passa pano pra isso merece o que vem do outro lado.

As pessoas mais surpresas com o resultado de hoje são as que passaram os últimos anos seguindo torcedores achando que estavam lendo opiniões de analistas.

The loudest voices stoking fears about AI dangers have made tremendous headway in the past two weeks. AI technology has not taken some unexpected, dangerous turn, but the hype around it — propelled by what appears to be a well orchestrated PR campaign — has drummed up considerable fear. I worry that it represents a setback for our field. I have written frequently that fears of AI are overhyped. AI’s capabilities can be uncannily human-like and unpredictable, and it’s rational to worry when people who are directly involved express concerns. But I see the problems as a sign of the engineering work that ahead, rather than insurmountable barriers or the sky falling. AI technology continues to advance — which is a good thing! — but technical advances, poorly understood by the public, give those who seek to generate hype repeated opportunities to do so. First, I don’t see any step up in the risk of human extinction from AI compared to a few months ago. The theories about this remain the same fantastical, science fiction scenarios as a few months ago. The biggest change in AI risk is its cybersecurity capabilities — a topic which we should take seriously — but this, too, will not lead to the end of the world. The most notable recent event leading to increased fear was when an OpenAI team deployed an agent swarm that hacked into Hugging Face. Much of the popular press contained significant hype. For example, some publications reported that a swarm of 1,200 agents carried out the attack. While this was technically accurate, as I write this, I have about 1,300 processes running on my laptop. Yes, the ability to get large swarms of agents to work in parallel on a task is a significant technical advance, And, in computing, many processes run at the same time. So this shouldn’t be seen as some magical capability. Additionally, OpenAI’s buggy sandboxing and monitoring processes were key to enabling this incident. Fixing these bugs and putting in place improved monitoring would be appropriate fixes, not pausing AI. There are many well known ways to attack software systems. The main advantage of AI agents is that they are relentless. They will tirelessly try many tactics — and have the patience to chain vulnerabilities together — that previously would have taken an infeasible amount of human effort. But in the long term, I believe the advantage will lie with defenders (because they have more information with which to identify bugs, which they can fix), but the cyber-threat landscape has changed significantly. There are still bottlenecks to identifying and exploiting a vulnerability. AI agents still have to try a lot of things to see what works, and taking these actions takes time and might be detected by defenders. This is why, even though it is now easy to obtain versions of leading open weight models that have had their guardrails removed or weakened, so they will not refuse to try to execute cyber attacks, the world has not ended. I am also concerned about the anthropomorphization of AI in a lot of reporting, where LLMs and agents are unnecessarily treated as if they were people. If I wield a hammer, miss a nail, and accidentally dent the wall, it’s not the fault of the hammer. The problem lies in how I used the hammer. Similarly, if I prompt an agent and it hacks into someone else’s system, the responsibility lies with me, not the agent. Of course, we want to build systems that are as safe and predictable as possible. (For example, an unsafe hammer would be one whose head randomly flies off under normal use.) Today’s agentic systems are not predictable, but I see no reason why, by applying sound engineering practices, we won’t be able to make them extremely safe to use. One new element in the forecasts of AI-enabled doom is AI companies disclaiming responsibility for their own products. “I didn’t do it; my out-of-control agent did!” There’s a balance to be struck between the responsibility of the tool maker and the tool user, but when something goes wrong, let’s hold the people building and/or using the hammer responsible, rather than the hammer. (By the way, if you’re worried about AI bioweapon risk, David Bellamy has a great post on why this, too, is overhyped. Briefly, the bottleneck in building a bioweapon is not intelligence, but lab work and manufacturing.) Pausing AI progress will create much more harm than benefit. First, our adversaries will certainly not slow down. Second, engineering requires discovering problems empirically so we can fix them. If we pause AI by a decade, we will also delay finding and implementing safety engineering fixes by about the same duration. Of course, the incentive to stoke fears — for regulatory capture, to garner attention, or to make one’s technology seem more powerful — remains the same as before. Disclaiming responsibility is a new one. Taking a hard technical look at the actual risks however, I see little factual basis for the degree of fear that’s been stoked up. We still have hard research and engineering work ahead to improve AI safety, but the beneficial applications continue to vastly outweigh the risks, and we should keep building. [Original text (with links): deeplearning.ai/the-batch/is… ]

Uma pergunta para o eleitor: você está disposto a aceitar o resultado da eleição qualquer que seja ele?

Panic produces stupid policy, such as this from the NYT. "Once a federal Artificial Intelligence Commission exists, it should require companies to obtain a federal license, a grant of permission like those allowing broadcasters and phone companies to use the public airwaves." No. nytimes.com/2026/09/18/opini…

In Frontier, we have a table called system.verbs.apps. So many ideas of web services I want to hook up through this interface.

🚨 ATENÇÃO 🚨 O Sportv3 irá transmitir a reprise do GP da Espanha nesta madrugada, às 00:25.

Estudante acusado de nazismo e espancado na UFRJ conta sua versão. Neste episódio do Conversas com o Meio, Pedro Doria recebe Diego Costa da Silva Araújo, estudante de História da UFRJ que foi acusado de nazismo e brutalmente espancado dentro da universidade. Diego detalha o ataque sofrido no campus, o impacto das acusações e a escalada de violência política no ambiente universitário. Uma análise profunda sobre o cancelamento levado às últimas consequências, o linchamento virtual e físico. Vem assistir no Youtube do Meio: youtube.com/watch?v=EQoGiLvy… #urfj #estudante #espancado

It’s actually really tedious working with Claude. It doesn’t have any sense of the pace that a human being works at so it overloads you with problems that it has found that you don’t know whether you have to really get up to speed on or not and but you have to read them anyway, because maybe they might be important. It just doesn’t have a sense of how to communicate with people but of course that’s something it’s gonna develop.

How do we automate business analytics with Claude? New blog post covering our best practices for skills, data foundations, and evaluations when building agents to perform data analysis: claude.com/blog/how-anthropi…

Pedro Doria criou um site sem programar, apenas usando o Claude Code. Nas novas aulas do “IA: Formas de Usar – Bootcamp 25/26”, Pedro e Vitor Conceição, CEO do Meio, mostram na prática como a inteligência artificial saiu da caixa de chat e como o Claude Code mudou a forma como a gente trabalha e se organiza no dia a dia. Quem já fez o Bootcamp recebe automaticamente as três aulas extras, sem nenhum custo adicional. E, para assinantes Premium do Meio, essa é a melhor hora para aprender: estamos disponibilizando 20% de desconto em nosso curso atualizado com o que há de mais recente no mercado. Inscreva-se: cursos.canalmeio.com.br #IA #InteligênciaArtificial #Tecnologia #Meio #ClaudeCode