More than half of researchers now use AI for peer review often against guidance https://t.co/4YlS1DlsAt @Nature

New paper from Apple - Sharp Monocular View Synthesis in Less than a Second Mescheder et al. @ Apple just released a very impressive paper (congrats! 🎉🥳). You give it an image and it generates a really great looking 3d Gaussian representation. Uses depth pro. It's really good. The model is about 3GB, and takes ~5-10s on my M1 Max per image. Single feedforward pass network. The video is using "Metal Splatter" to view the .ply from ml-sharp on Apple Vision Pro. Some really wow moments today trying it on different scenes.

@graykevinb @AndreTI @chris_j_paxton There are a few chargers there. https://t.co/AXv5moEIQu

A one step solution to crime: Lock up the criminals. It worked in El Salvador: https://t.co/S5jYtBOpMO

BREAKING: Grokipedia just crossed 54,000 approved edits by Grok and now hosts over 1 million articles. https://t.co/irjFuVghkD

Underground City on Mars https://t.co/WK2cpBBwXl

NEWS: Starlink’s global internet traffic more than doubled in 2025, based on new data from Cloudflare. It now has ~8 million active users in more than 150 markets, up from around 4 million in 2024. https://t.co/OfFWUZkUQf

Elon Musk: The amount of indoctrination that's happening in schools and universities is far beyond what parents realize --- “The amount of indoctrination that's happening in schools and universities is, I think, far beyond what parents realize. Let me give you an example that a friend of mine told me. His daughters go to high school in the Bay Area. And he was asking them, like, 'You know, who are the first few presidents of the United States?' They could name Washington. Then he said, 'Well, what do you know about him? Well, 'He was a slave owner.' 'What else?' Like, OK, that's, maybe you should know more than that. Slavery is obviously a horrific institution, but we should still know more about George Washington than that.” Source: @billmaher , @elonmusk

Building “RAG 2.0” is just making Claude Code running over your filesystem 🤖🗂️ To make this work well, you need to solve three things 1️⃣ Virtualize your filesystem to prevent the agent from messing stuff up. AgentFS by @tursodatabase is a nice example of how you can give the agent access to a copy of all your files without messing up your raw data. 2️⃣ Parse unstructured documents like PDFs, pptx, Word into an LLM-ready format. Agentic OCR solutions like LlamaParse can help here 3️⃣ Creating an agentic loop with human-in-the-loop. If you want to control the agent implementation instead of using Claude Code out of the box, you can use @llama_index workflows to help orchestrate these long-running agent tasks. Shoutout @itsclelia, check it out! Blog: https://t.co/29DGNFmMQl Repo: https://t.co/rngUWQHoqZ

+1 more

生成 AI は「人の成長を加速させるパートナー」 鼎談：https://t.co/s6aKwqkg50 生成 AI をどのように活用すれば、新たな価値を生み出すような成果を得られる？3 名の鼎談を通じて、生成 AI の役割を紐解いていきます。 🔵Sakana AI デイビッド・ハ氏 🟡ビズリーチ 外山 英幸氏 🟢Google Cloud 邑本 卓也

生成 AI は「人の成長を加速させるパートナー」 鼎談：https://t.co/s6aKwqkg50 生成 AI をどのように活用すれば、新たな価値を生み出すような成果を得られる？3 名の鼎談を通じて、生成 AI の役割を紐解いていきます。 🔵Sakana AI デイビッド・ハ氏 🟡ビズリーチ 外山 英幸氏 🟢Google Cloud 邑本 卓也

introducing Ariana Automations every other tool gives agents a sandbox. we give them a full linux computer with sudo, docker, their own IP now you can run scripts based on what the agent does setup the environment once, then get rid of the human in the loop (you) a 🧵 https://t.co/rMvnXkgGCn

That's my somewhat-romantic intuition too https://t.co/AuNLgq0DI5

That's my somewhat-romantic intuition too https://t.co/AuNLgq0DI5

opens with one of the all time epigraphs https://t.co/eRcT6UbjOC

opens with one of the all time epigraphs https://t.co/eRcT6UbjOC

The Final Reward credit: @snwy_me https://t.co/YjJBjXEM1F

Introducing VL-JEPA: Vision-Language Joint Embedding Predictive Architecture for streaming, live action recognition, retrieval, VQA, and classification tasks with better performance and higher efficiency than large VLMs. • VL-JEPA is the first non-generative model that can perform general-domain vision-language tasks in real-time, built on a joint embedding predictive architecture. • We demonstrate in controlled experiments that VL-JEPA, trained with latent space embedding prediction, outperforms VLMs that rely on data space token prediction. • We show that VL-JEPA delivers significant efficiency gains over VLMs for online video streaming applications, thanks to its non-autoregressive design and native support for selective decoding. • We highlight that our VL-JEPA model, with an unified model architecture, can effectively handle a wide range of classification, retrieval, and VQA tasks at the same time. by @Delong0_0 @MustafaShukor1 @TheoMoutakanni @willyhcchung Jade Lei Yu Tejaswi Kasarla @AllenBolourchi @ylecun @pascalefung https://t.co/oUnjCaMKVv

I interviewed @ylecun with @ziv_ravid! https://t.co/R83gdbBg9h

I interviewed @ylecun with @ziv_ravid! https://t.co/R83gdbBg9h

📁 Yann LeCun explains that LLMs work well when problems are symbolic, like math, code or chess, where searching through known sequences is enough. But the real world does not work that way. Physical action, planning and understanding what is possible require continuous intuition, which LLMs lack. Manipulating symbols is not the same as understanding reality.

Yann LeCun explains that large language models are trained on about 30 trillion words, representing nearly all public internet text. He says it would take a human over 500,000 years to read that much. But a 4-year-old child sees just as much visual data in their first few years of life. This shows how much richer and more complex real-world experience is compared to reading text. Training on the web is huge but it still doesn’t match what a child learns just by living.

Yann LeCun warns that in the near future, all our digital interactions will be filtered through AI assistants. That means our entire "information diet" what we read, see, and hear, could be controlled by just a few powerful tech companies. He compares this to the media: just like we need diverse press, we need diverse AI systems. If we don’t build open platforms, a few companies could control global information flow. This is his biggest fear. Not AI going rogue, but AI being monopolized.

📢Manusアップデートのお知らせ 最新アーキテクチャシリーズ Manus 1.6（Max~Lite）を全ユーザー向けにリリースしました！ 🔹最高峰のパフォーマンス (Max) └ 最高性能の基盤モデルと刷新されたエージェントアーキテクチャにより、「より粘り強く、より確度の高い」タスク実行力を手に入れました。また、Maxはエクセル操作に非常に優れており、グラフや複雑な関数を含む複数シートのエクセルをワンプロンプトで生成可能です。 🔹モバイルアプリ開発 └ 進化したWebDev機能に加えて、Manus上でスマホアプリ開発〜APK出力まで出来るようになりました。「アプリ化」をより気軽に、より高速に。Manusでアイデアを形にする体験をサポートします。 ※ ストア公開機能は今後のアップデートで正式サポートを予定しています 🔹直感的な画像デザイン └ Nano Banana Pro画像のテキスト編集がついに可能になりました。新たに追加されたDesign View（Canvas）機能では、生成した画像のテキスト抽出、編集、拡大、背景削除まで、一つのインターフェースで完結できます。 また、ローカルからの画像ファイルのアップロードから編集を行うことも可能です。

📢Manusアップデートのお知らせ 最新アーキテクチャシリーズ Manus 1.6（Max~Lite）を全ユーザー向けにリリースしました！ 🔹最高峰のパフォーマンス (Max) └ 最高性能の基盤モデルと刷新されたエージェントアーキテクチャにより、「より粘り強く、より確度の高い」タスク実行力を手に入れました。また、Maxはエクセル操作に非常に優れており、グラフや複雑な関数を含む複数シートのエクセルをワンプロンプトで生成可能です。 🔹モバイルアプリ開発 └ 進化したWebDev機能に加えて、Manus上でスマホアプリ開発〜APK出力まで出来るようになりました。「アプリ化」をより気軽に、より高速に。Manusでアイデアを形にする体験をサポートします。 ※ ストア公開機能は今後のアップデートで正式サポートを予定しています 🔹直感的な画像デザイン └ Nano Banana Pro画像のテキスト編集がついに可能になりました。新たに追加されたDesign View（Canvas）機能では、生成した画像のテキスト抽出、編集、拡大、背景削除まで、一つのインターフェースで完結できます。 また、ローカルからの画像ファイルのアップロードから編集を行うことも可能です。

El Salvador and @xai will redefine the future of public education. https://t.co/9ZsKT8cOV0

+2 more

bro https://t.co/mtOaAPCEDl

bro https://t.co/mtOaAPCEDl

@youwouldntpost https://t.co/ai9wpXtM1k

@youwouldntpost https://t.co/ai9wpXtM1k