magicpro @MAGICPRO - Twitter Profile

イギリスでカツカレーが流行り、カツが抜けてもカツカレー呼びされ，その上「カツ＝カレー味」の意味になった、と聞くと「そもそもカツとはお前の国のコートレットの事だ」と言いたくもなる。だが省略された粘液状の「カレー」のことを今のインドでは「グレービー」と英語くさい呼び方をするらしい（具や肉が入るとマサラ）と聞いて、もうなにをどう説明したのわかんねえな、という顔になっている

41

8K

3K

764

631K

magicpro @magicpro

4 days ago

ポイント: It avoids controversy by talking about anime, sushi, and tradition.

Superman

@thesupermanmx

6 days ago

Researchers proved every major LLM is secretly obsessed with Japan. And they finally figured out why. For years, we’ve been told that AI is entirely Western-centric, that it just reflects Silicon Valley and American values. A landmark paper by Cardiff and Basque researchers tested 31,680 cultural prompts across 24 languages on frontier models like ChatGPT, Claude, and Gemini. The results shattered that assumption. In six out of eight frontier models, Japan was the single most frequently referenced country when asked open-ended cultural questions. Ask about traditional dances, festivals, or everyday practices in an open context, and the AI defaults to Japan. Over and over again. Here is the twist nobody expected. This bias doesn't come from raw pre-training internet data. The researchers tracked where the obsession forms. It emerges after pre-training, during the supervised fine-tuning and alignment phase when humans teach the AI how to behave. Why Japan? Because decades of global soft power, rich cultural export, and clean, universally admired digital archives make Japanese culture uniquely "safe" for AI safety filters to lean on. When labs train models to be harmless and universally pleasing, the AI defaults to the cultural equivalent of comfort food. It avoids controversy by talking about anime, sushi, and tradition.

thesupermanmx's tweet photo. Researchers proved every major LLM is secretly obsessed with Japan.

And they finally figured out why.

For years, we’ve been told that AI is entirely Western-centric, that it just reflects Silicon Valley and American values.

A landmark paper by Cardiff and Basque researchers tested 31,680 cultural prompts across 24 languages on frontier models like ChatGPT, Claude, and Gemini.

The results shattered that assumption.

In six out of eight frontier models, Japan was the single most frequently referenced country when asked open-ended cultural questions.

Ask about traditional dances, festivals, or everyday practices in an open context, and the AI defaults to Japan.

Over and over again.

Here is the twist nobody expected.

This bias doesn't come from raw pre-training internet data.

The researchers tracked where the obsession forms. It emerges after pre-training, during the supervised fine-tuning and alignment phase when humans teach the AI how to behave.

Why Japan?

Because decades of global soft power, rich cultural export, and clean, universally admired digital archives make Japanese culture uniquely "safe" for AI safety filters to lean on.

When labs train models to be harmless and universally pleasing, the AI defaults to the cultural equivalent of comfort food.

It avoids controversy by talking about anime, sushi, and tradition.

230

10K

2K

5K

4M

0

94

magicpro retweeted

ねこみみパラオ🇻🇦🇯🇵✝️ @nekomimipalau

7 days ago

おたくの国から呼んだ技術者が助けてくれたんですよ、それ

19

14K

688

887

1M

magicpro @magicpro

6 days ago

夏にスーツを着るのは有罪ってことにしよう

ジロウ

@jiro6663

6 days ago

日中に買い物行くのが危険な気候になってくると東南アジアの街々でナイトマーケットが盛んなのがよくわかるようになってきた。日本も危険な暑さに適応した生活文化を再編成する必要があるよなあ

35

11K

3K

459

573K

0

214

magicpro @magicpro

6 days ago

ロメールの映画はゆったりと観るものはまったく同感

DepressedBergman

@DannyDrinksWine

7 days ago

Éric Rohmer explains why he didn't like his movies being screened at crowded theaters: "If it depended only on me, instead of attracting people to my films, I would try to drive them away. I would tell them the films are more difficult than they really are, because I don’t like to deceive people, I like to show my films to people who can appreciate them. I’m not interested in the number of spectators. Having said that, it’s true that a film is a commercial undertaking and ought to recover its costs. But as my films don’t cost much, I don’t think I need a very large audience, and I’ve always thought that they should be shown in theaters that aren’t too big. The intimate character of my films doesn’t suit a theater or an audience too large for them. And I don’t think they are suited to a mass reaction or a collective reaction. It’s better if the spectator feels he is experiencing a completely personal reaction to it. Each reaction should be unique, individual, and different. I think the film is enjoyed better if the spectators aren’t sitting too near one another, if the theater isn’t too full, and they don’t know each other. Then each has a different reaction. That’s better than a theater where there’s a uniform reaction. I don’t like watching one of my films in public and it distresses me if everyone laughs in the same place, as my film wasn’t made with that in mind. I didn’t write something just to make everyone laugh at the same time. It’s all right if someone smiles, but it shouldn’t happen at exactly the same place in the film. Perhaps this is because my films are more like reading than like watching a spectacle; they are made more to be read like a book than seen like something on the stage. So it distresses me to see a collective reaction." (Éric Rohmer's interview with Éric Rohmer, 1971) Clip from: La collectionneuse (1967) Director: Éric Rohmer