App Store、ボリすぎ

来週Grok 3.5がSuperGrokユーザーを対象にリリースされるみたい

Threadsって自分の事を可愛いと思っている女子がセルフィーを投稿するプラットフォームなんですよね？

アコギが欲しい

ヨドバシカメラのNintendo Switch 2抽選販売にも申し込んでみるか

AmazonでNintendo Switch 2の招待リクエストをしてみた

おはようございます

遅ればせながらClaudeにもリサーチ機能が実装されたけど、レポートの質はChatGPTやGeminiの方が上という印象。

Google Geminiも高額なサブスクリプションプランを計画しているらしい。いくらになるんだろう。

Soon, users on the Claude Max plan will be able to link their accounts with Claude Code and use it under the same usage limits.

I’ll never use the Meta AI app

Anthropic CEO Dario Amodei admits AI companies don't understand how their systems make decisions, comparing them to unpredictably growing organisms rather than traditional programmed software. He aims for Anthropic to detect most AI model problems by 2027 through increased research, government oversight, and prioritizing safety measures.

WebProNews

Anthropic CEO Admits the Company Does Not Understand How AI Works

Anthropic CEO Dario Amodei has made a disturbing admission, saying his company doesn't understand how the AI models it is developing actually ...

Wtf, the Intel-powered Surface Laptop 7 is expensive!

What I don't like about Gemini 2.5 Pro is that it doesn't search the internet to retrieve the latest information unless I say so. On the contrast, Claude 3.7 Sonnet automatically searches the internet when the model doesn't have answers.

The M4 Max Mac Studio is incredibly fast and justifies its high price with substantial workflow enhancements, particularly in video and audio editing. It’s lauded for its quiet operation and generous number of ports, although the rear headphone jack and upgrade costs are seen as drawbacks. Some users also mentioned the headphone jack issue but commended the built-in audio quality. Overall rating: 4.5/5.

AppleInsider Forums

Mac Studio M4 Max review one month later: Costly computing power, worth every cent - Current Mac Hardware Discussions on AppleInsider Forums

Switching to an M4 Max Mac Studio was the most expensive Apple purchase I

Tests show that newer ChatGPT models hallucinate significantly more than older versions. The o3 model has a 33% hallucination rate, while the o4-mini hallucinates 48% of the time, compared to previous models (o1, o3-mini) which had 14-16% rates. OpenAI doesn't know why it's happening TechCrunch and says more research is needed.

Digital Trends

It’s not your imagination — ChatGPT models actually do hallucinate more now

According to internal tests, newer models like o3 and o4-mini hallucinate significantly more than older versions, and OpenAI doesn't know why.

Apple will completely rebuild Siri's architecture in iOS 19, merging its current dual-system approach (one for basic commands, another for advanced queries) into a unified "LLM Siri" architecture. This overhaul, expected to launch by spring 2026, will enable the delayed features like personal context and on-screen awareness, whilst significant leadership changes, including Mike Rockwell taking charge of Siri development, signal Apple's serious commitment to finally addressing Siri's longstanding issues.​​​​​​​​​​​​​​​​

9to5Mac

Apple wants to fix Siri in iOS 19, here's how - 9to5Mac

It goes without saying, Apple’s approach to developing its AI-infused version of Siri has gone very wrong in many ways....

AppleがついにSiriの根本的な欠陥を認め、iOS 19で徹底的な作り直しを行うことで、長年の問題を解決しようとしているようです。 現在のSiriが「二つの脳」という非効率的な構造を持っている問題を解消するため、AppleはSiriのアーキテクチャを一から再構築し、遅延していた重要な機能を実現させようとしています。これは単なる機能追加ではなく、Siriを真に有用なアシスタントへと進化させるための本質的な取り組みです。

9to5Mac

Apple wants to fix Siri in iOS 19, here's how - 9to5Mac

It goes without saying, Apple’s approach to developing its AI-infused version of Siri has gone very wrong in many ways....

Nintendo Switch 2が欲しい。抽選に申し込んですらいないけど。

DeepSeek R2's specifications: - Unit cost reduced by 97.3%, ready for immediate release - Its self-developed distributed training framework achieves 82% utilization of 910B chip clusters - Reaches 512 PetaFLOPS computing power under FP16 precision - Achieves 91% efficiency compared to same-scale A100 clusters (data verified by Huawei laboratory)