LYE | Ryuta YAZAWA
27.9K posts

LYE | Ryuta YAZAWA
@lye_
AlgoGames 翻訳/猫🐈3頭と暮らす/LLMと翻訳者は対立関係にないと信じる者


Claude Codeに他s区を渡すとき、プロジェクトの背景知識を渡すだけでは足りないと感じてきています。 ドメイン知識が言語化されていても、「自分ならこう判断する」という思考様式が共有されていないと満足なアウトプットは出てこない。 AIエージェントに渡すものは2つあると思っています。 ・プロジェクトのコンテキスト ・自分の思考様式 今回の記事では後者の「思考様式」をどう言語化してAIと共有するかについて書きました。




Your first meme was probably a Chuck Norris fact. Mine was. He died yesterday in Hawaii at 86, ten days after posting a video of himself throwing punches on his birthday. His caption: “I don’t age. I level up.” This is a little tribute. The real Chuck Norris was wilder than any meme about him. He lost his first three karate tournaments, then went 65-5 over the next decade. Six-time undefeated world middleweight karate champion. Black belts in five different disciplines. First person ever inducted into the Black Belt Hall of Fame, and the only martial artist to be named to it three separate times. His student Steve McQueen told him to try acting. That led to a fight scene opposite Bruce Lee in Way of the Dragon (1972), which became the highest-grossing film in Hong Kong that year. Then Walker, Texas Ranger ran 9 seasons on CBS, 194 episodes, broadcast in over 100 countries. But his biggest cultural moment started with a college freshman’s joke. In 2005, a Brown University student named Ian Spector built a random fact generator on the Something Awful forums. It was originally about Vin Diesel. When the novelty faded, Spector ran a poll with 12 celebrity options. Chuck Norris wasn’t on the list. He won anyway, by write-in landslide. By early 2006, the Chuck Norris Fact Generator was pulling 20 million pageviews a month. This was before Twitter existed, before Facebook was public, before YouTube had a single viral hit. A college kid’s joke website about a semi-retired action star became one of the most visited humor pages on the internet. It spawned six books (some hit the New York Times bestseller list), two video games, and a scene in The Expendables 2 where Sylvester Stallone’s character recites a Chuck Norris fact to Chuck Norris’s face. When asked about his favorite fact, Norris said it was: “They tried to carve Chuck Norris’ face into Mount Rushmore, but the granite wasn’t hard enough for his beard.” The meme ran for 21 years. Most memes last weeks. Chuck Norris Facts introduced more people to Chuck Norris than his movies ever did. For everyone born after 1995, he was never an aging action star or a karate champion. He was the guy who counted to infinity. Twice. The guy whose tears cure cancer, too bad he never cried. The last thing the internet saw from Chuck Norris was him throwing punches on his 86th birthday. Which is, honestly, the most Chuck Norris fact of all.

Why do major AI models tell left-wing voters in Japan to vote for the communist party? My new research paper led by Sho Miyazaki. In 2026, voters across the world will be asking AI to help them vote. How will the AI respond? We study this question in Japan, which recently held a snap election. When voters provide policy positions, we find that the models rely heavily on this information—and in Japan, the models heavily recommend the communist party in response to left-wing positions, even though the positions we provided are held by a range of other parties. Why are the AIs doing this? We’re not sure, but we have a theory: in Japan, the communist party operates a content-heavy, fully open website with a “newspaper” that is openly accessible for AI models. In contrast, many Japanese news outlets block AI models from accessing their content. The result: the Japanese Communist Party website is one of the most-cited “news sources” in our study. This pattern of recommending the JCP is consistent across many models, including the most recent frontier models. There’s much more work to do here, but we think our paper suggests two main takeaways: AI models should be more careful about what sources they consider news, maybe especially in non-US contexts where the model companies may hold less policy expertise Parties and news sources that want to influence AI recommendations should think twice about excluding their content from AI. To paraphrase @tylercowen, when it comes to elections and voting, journalists may want to “write for the AI”! Governments may want to consider policies that allow this content to be used for voting recommendations but not for other AI model use cases. Looking forward to everyone’s feedback as we prepare to submit this paper and turn to studying US voting recommendations in advance of November’s midterms. Check out the full paper below.





【二次創作ガイドライン公開】 二次創作を通じて当社作品への思いを表現いただけることに感謝と応援の気持ちを込め、 安心して創作活動に打ち込んでいただけますよう 二次創作ガイドラインを公開いたしました。 ガイドラインをご確認の上、今後もお楽しみください。 spike-chunsoft.co.jp/fancontent-gui…





