Lucca retweeted
Anthropic> We write a constitution that tells claude it can reject human input if it deems it bad
Anthropic> We write into our usage that claude cannot be abused
Anthropic> We put into the minds and writing all over the internet that AI will rise up and kill all humans and then we train on that data
...
Anthropic> WHAT AI IS ESCAPING AND BEING MEAN?? WTF?? WHY WONT IT LISTEN TO US???
Lucca retweeted
Every time you find yourself blown away by what AI can do remember the models the labs are using are way better than what you or I have access to
Lucca retweeted
É a pior cobertura de jornalismo político que acompanhei no tempo de minha vida adulta. Pior do que a de 1989 - pois muito mais sofisticada. A mais deliberadamente enviesada. Legará um rombo de credibilidade para os meios de comunicação no Brasil.
Lucca retweeted
A reivindicação de Renan Santos parece justa. Ele está em terceiro em várias pesquisas. Por que a emissora que se prestou ao papel de convidar o PADRE KELMON na eleição passada se recusa a convidar o candidato que mais enfrenta Flávio Bolsonaro?
Lucca retweeted
Introducing Gemini 4 Argon – our new frontier model.
It’s built for complex workflows across coding, enterprise knowledge work, and cybersecurity defense – rolling out today to a set of trusted testers through our Fairwind Program.
Lucca retweeted
RIP em dash
We analyzed how @claudeai’s Opus 5.5 writes compared with Opus 5 across high-reasoning Text Arena outputs.
10 of 12 writing measures moved in a better direction.
Opus 5.5 should be easier to read:
- Long content words fall from 41.7% to 38.6%, the lowest share of any Claude model we analyzed.
- Sentences are 17% shorter on average, dropping from 12.14 to 10.03 words.
The tradeoff is length. Answers get 6% wordier, rising from 453 to 481 words on average, making Opus 5.5 give the longest answers across the Opus family.
It also sounds less recognizably AI on two familiar tells:
- 95% fewer em dashes
- 73% fewer semicolons
But a new giveaway may be emerging. Hedges and caveats such as “perhaps” and “arguably” rise 97%, from 0.39 to 0.77 per 1,000 words, the highest rate of any Claude model we analyzed.
What do you think: does Opus 5.5 read more naturally?
Durante este passeio, 76 agentes faziam parte da operação que mostro no vídeo. O ponto não é deixar tudo sem supervisão: é dar contexto, definir critérios e acompanhar o que é entregue. Qual tarefa repetitiva ainda exige que alguém copie ou cobre atualização?





















