Crossing the uncanny valley of conversational voice
How do we know when someone truly understands us? It is rarely just our words—it is in the subtleties of voice: the rising excitement, the thoughtful pause, the warm reassurance. Voice is our most intimate medium as humans, carrying layers of meaning through countless variations in tone, pitch, rhythm, and emotion. Today’s digital voice assistants lack essential qualities to make them truly useful
Whisk: é ã«ã‚るアイデアをパッã¨ã‹ãŸã¡ã«
本日よりã€Google Labs ã§ç”Ÿæˆ AI を活用ã—ãŸç”»åƒç”Ÿæˆã®è©¦é¨“é‹ç”¨æ©Ÿèƒ½ Whisk を日本をå«ã‚€ 100 以上ã®å›½ã¨åœ°åŸŸã§é †æ¬¡æä¾›ã—ã¾ã™ã€‚Whisk ã¯ã€è©³ç´°ãªãƒ†ã‚スト プãƒãƒ³ãƒ—トã®å¿…è¦ãªãã€ç”»åƒã‚’プãƒãƒ³ãƒ—トã¨ã—ã¦ä½¿ç”¨ã™ã‚‹ç”»åƒç”Ÿæˆ AI ã§ã™ã€‚ç”»åƒã‚’é¸æŠžã—ã¦ã€Whisk ã«å…¥åŠ›ã™ã‚‹ã ã‘ã§ç°¡å˜ã«å§‹ã‚ã‚‹ã“ã¨ãŒã§ãã¾ã™ã€‚ （一部ã®æ©Ÿèƒ½ã«ã¤ã„ã¦ã¯ã€Whiskを最大é™ã«æ´»ç”¨ã—ã¦ã„ãŸã ããŸã‚ã€è‹±èªžã§ã®ã”利用を推奨ã—ã¦ãŠã‚Šã¾ã™ï¼‰ Whisk ã¯ã€ãƒ¢ãƒ‡ãƒ«ã€ã‚·ãƒ¼ãƒ³ã€ã‚¹ã‚¿ã‚¤ãƒ«ã”ã¨ã«è‡ªåˆ†ã®å¥½ããªç”»åƒã‚’入力ã§ãã¾ã™*。入力後ã€ãれらを組ã¿åˆã‚ã›ã¦ã€é¸æŠžã—ãŸã¬ã„ãã‚‹ã¿ãªã©ã®ãƒ†ãƒ³ãƒ—レートã«åˆã‚ã›ãŸã‚ªãƒªã‚¸ãƒŠãƒ«ã®ç”»åƒã‚’生æˆã§ãã¾ã™ã€‚ã¾ãŸã€ä»Šå›žã®æ—¥æœ¬ã§ã®æä¾›ã«ã‚ã‚ã›ã¦ã€æ—¥æœ¬æ–‡åŒ–ã‹ã‚‰ã‚¤ãƒ³ã‚¹ãƒ‘イアã•ã‚ŒãŸæ–°ã—ã„ï¼’ã¤ã®ãƒ†ãƒ³ãƒ—レートã€ã€Œã‚«ãƒ—セルトイã€ã¨ã€ŒãŠå¼å½“ã€ã‚‚æ–°ã—ãè¿½åŠ ã—ã¾ã—ãŸã€‚ã¾ãŸã€ä»Šå›žã¯ãƒãƒ¬ãƒ³ã‚¿ã‚¤ãƒ³ãƒ‡ãƒ¼ã«åˆã‚ã›ãŸç‰¹
Dario Amodei — Machines of Loving Grace
Machines of Loving Grace1 How AI Could Transform the World for the Better October 2024 I think and talk a lot about the risks of powerful AI. The company I’m the CEO of, Anthropic, does a lot of research on how to reduce these risks. Because of this, people sometimes draw the conclusion that I’m a pessimist or “doomer†who thinks AI will be mostly bad or dangerous. I don’t think that at all. In fa
Build an MCP Server · Cloudflare Agents
Model Context Protocol (MCP) ↗ is an open standard that allows AI assistants & LLMs to interact with services directly. If you want users to access your service or product straight from their AI assistant, you can enable this by spinning up an MCP server for your application. Normally, setting up an MCP server requires writing boilerplate code to handle routing, define types, and standing up a ser
GitHub - deepseek-ai/DeepSeek-R1
We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrated remarkable performance on reasoning. With RL, DeepSeek-R1-Zero naturally emerged with numerous powerful and interesting reasoning behaviors. However, DeepSeek-R1-Z
Dify.AI · 先進的ãªAIアプリケーションã®ãŸã‚ã®ã‚¤ãƒŽãƒ™ãƒ¼ã‚·ãƒ§ãƒ³ã‚¨ãƒ³ã‚¸ãƒ³
Difyã¯ã‚ªãƒ¼ãƒ—ンソースã®LLMアプリ開発プラットフォームã§ã™ã€‚RAGエンジンを使用ã—ã¦ã€ã‚¨ãƒ¼ã‚¸ã‚§ãƒ³ãƒˆã‹ã‚‰è¤‡é›‘ãªAIワークフãƒãƒ¼ã¾ã§LLMアプリを編æˆã—ã¾ã™ã€‚
AI編集者ã¨ã¤ãã‚ã†ï¼ã‚ãŸã—ã®ç¾ä»£æ–°æ›¸ï½œè¬›è«‡ç¤¾ ç¾ä»£æ–°æ›¸ 創刊60周年
60周年を記念ã—ã¦åˆ¶ä½œã—ãŸã‚µã‚³ãƒƒã‚·ãƒ¥ãƒ»å†Šå・å“上カレンダー・ステッカーã®ã‚»ãƒƒãƒˆãŒã€ãƒ•ã‚©ãƒãƒ¼ï¼†ãƒã‚¹ãƒˆã§10å様ã«å½“ãŸã‚‹ï¼ 応募方法 講談社ç¾ä»£æ–°æ›¸å…¬å¼X（@gendai_shinsho）をフォãƒãƒ¼ã—ã¦ã€ã‚ãŸã—ã®ç¾ä»£æ–°æ›¸ã§ã¤ãã£ãŸã‚ãªãŸã ã‘ã®æœ¬ã‚’ã€ä»¥ä¸‹ã®2ã¤ã®ãƒãƒƒã‚·ãƒ¥ã‚¿ã‚° #ç¾ä»£æ–°æ›¸60周年ã‚ャンペーン#ã‚ãŸã—ã®ç¾ä»£æ–°æ›¸ã‚’ã¤ã‘ã¦Xã«ãƒã‚¹ãƒˆã—よã†ï¼
Introducing OpenAI Japan
As we grow our operations internationally, we’re expanding into Asia with a new office in Tokyo, Japan. We are committed to collaborating with the Japanese government, local businesses, and research institutions to develop safe AI tools that serve Japan’s unique needs and to unlock new opportunities. We chose Tokyo as our first Asian office for its global leadership in technology, culture of servi
ãªãœæˆ‘々㯠GitHub Copilot Enterprise ã®å°Žå…¥ã‚’見é€ã£ãŸã®ã‹ - 一休.com Developers Blog
CTO 室ã®æ©ç”°ã§ã™ã€‚ 今回㯠GitHub Copilot Enterprise を評価ã—ã¦ã¿ã¦ã€ç¾æ™‚点ã§ã¯ã¾ã 採用ã—ãªã„ã“ã¨ã‚’決ã‚ãŸã€ã¨ã„ã†ãŠè©±ã‚’ã”紹介ã—ãŸã„ã¨æ€ã„ã¾ã™ã€‚ ãã£ã‹ã‘ ã¨ã‚るエンジニア㌠Slack ã§è‡ªèº«ã® times ãƒãƒ£ãƒãƒ«ã«æ™‚é›¨å ‚ã•ã‚“ã® GitHub Copilot Enterprise ã®ã‚¹ã‚¹ãƒ¡ã¨ã„ã†è¨˜äº‹ã‚’投稿ã—ãŸã“ã¨ãŒç™ºç«¯ã§ã—ãŸã€‚特ã«æ„Ÿæƒ³ã¯ãªã URL 㫠👀 ã ã‘ãŒæ·»ãˆã‚‰ã‚Œã¦ã„ãŸã®ã§ã€å¾Œã§è¦‹ã‚‹ãらã„ã®ãƒ¡ãƒ¢ã ã£ãŸã‚“ã ã¨æ€ã„ã¾ã™ã€‚ ãれを見ã‹ã‘ãŸåˆ¥ã®ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ãŒæŠ€è¡“雑談ãƒãƒ£ãƒãƒ«ã«ãã®æŠ•ç¨¿ã‚’共有ã—ã¦ã€ã“ã‚Œã¯å‡„ãã†ã¨è©±é¡Œã‚’å‘ã‘ãŸã¨ã“ã‚ã€CTO ã®ã€Œè©•ä¾¡ã—ã¦ã¿ã‚‹ï¼Ÿã€ã®ä¸€è¨€ã§ã€æœ‰å¿—ãŒé›†ã£ã¦è©•ä¾¡ãƒ—ãƒã‚¸ã‚§ã‚¯ãƒˆãŒå§‹ã¾ã‚Šã¾ã—ãŸã€‚ 雑談ãƒãƒ£ãƒãƒ«ã§ãã£ã‹ã‘ã¨ãªã‚‹æŠ•ç¨¿ãŒå…±æœ‰ã•ã‚Œã¦ã‹ã‚‰ã€30分足らãšã®å‡ºæ¥äº‹ã§ã—ãŸï¼ˆç¬‘）。 ã“ã®è©±é¡ŒãŒå‡ºãŸã®ã¯é‡‘曜日ã§ã—ãŸãŒã€é€±æ˜Žã‘早々ã«ç¨Ÿè°ã‚’終ãˆã€ç«æ›œ
radical-pilot-321ã®ãƒ–ックマーク - ã¯ã¦ãªãƒ–ックマーク
EVã®ç§»è¡Œã‚¹ãƒ”ードãŒæ¸›é€Ÿã—ã¦ãã¦ã€ãƒã‚¤ãƒ–リッドãŒå†æ³¨ç›®ã•ã‚Œã¦ã„ã‚‹ãªã‹ã€ã‚¹ãƒãƒ«ã¯ãƒˆãƒ¨ã‚¿ã®ãƒ¢ãƒ¼ã‚¿ãƒ¼ï¼ç™ºé›»æ©Ÿã¨ä»•çµ„ã¿ã‚’使ã£ãŸã‚¹ãƒˆãƒãƒ³ã‚°ãƒã‚¤ãƒ–リッドを新開発ã—ãŸã€‚ãã®æ–°ã—ã„ストãƒãƒ³ã‚°ãƒã‚¤ãƒ–リッドã®é§†å‹•ç”¨ã¨ç™ºé›»ç”¨2ã¤ã®ãƒ¢ãƒ¼ã‚¿ãƒ¼ã€ãƒ•ãƒãƒ³ãƒˆãƒ‡ãƒ•ã€AWDã‚’1パッケージã«ã—ãŸã€Œãƒˆãƒ©ãƒ³ã‚¹ã‚¢ã‚¯ã‚¹ãƒ«ã€ã‚’生産ã—ã¦ã„るスãƒãƒ«åŒ—æœ¬å·¥å ´ã‚’è¦‹å¦ã™ã‚‹æ©Ÿä¼šãŒã‚ã£ãŸã€‚ã¯ãŸã—ã¦å·¥å ´ã‚’見å¦ã—ã¦ã©ã‚“ãªã“ã¨ã‚’æ„Ÿã˜ãŸã®ã‹ï¼Ÿ 文：国沢光å®ï¼å†™çœŸï¼šãƒ™ã‚¹ãƒˆã‚«ãƒ¼Web編集部ã€ã‚¹ãƒãƒ« ã€ç”»åƒã‚®ãƒ£ãƒ©ãƒªãƒ¼ã€‘スãƒãƒ«ã®ã‚¹ãƒˆãƒãƒ³ã‚°ãƒã‚¤ãƒ–リッドã¯ã©ã‚Œã»ã©å‡„ã„ã®ã‹ï¼Ÿç”Ÿç”£å·¥ç¨‹ã‚’写真ã§ãƒã‚§ãƒƒã‚¯ï¼ï¼ˆ47枚）画åƒã‚®ãƒ£ãƒ©ãƒªãƒ¼
1ビットLLMã®è¡æ’ƒ! 70Bã§8.9å€é«˜é€Ÿã€€å…¨ã¦ã®æŽ¨è«–ã‚’åŠ ç®—ã®ã¿ã§!GPUä¸è¦ã«ãªã‚‹å¯èƒ½æ€§ã‚‚
1ビットLLMã®è¡æ’ƒ! 70Bã§8.9å€é«˜é€Ÿã€€å…¨ã¦ã®æŽ¨è«–ã‚’åŠ ç®—ã®ã¿ã§!GPUä¸è¦ã«ãªã‚‹å¯èƒ½æ€§ã‚‚ 2024.02.28 Updated by Ryo Shimizu on February 28, 2024, 16:46 pm JST 2月ã¯ä¸å›½ã§ã¯æ˜¥ç¯€ã¨ã„ã†ãŠæ£æœˆãŒã‚ã‚Šã€æ˜¥ç¯€ã«ã¯ã¿ã‚“ãªä¼‘む。 ãれもã‚ã£ã¦ã‹ã€12月ã‹ã‚‰1月ã«ã‹ã‘ã¦æ€’涛ã®è«–文発表ãŒè¡Œã‚ã‚Œã¦æ¯Žæ—¥ã€Œãƒ‡ã‚¤ãƒªãƒ¼AIニュースã€ã‚’é…ä¿¡ã—ã¦ã„ã‚‹ç†è€…ã¯å¿™æ®ºã•ã‚Œã¦ã„ãŸã€‚ 春節ä¸ã«ã¯OpenAIãŒSoraã‚’ã€GoogleãŒGemini1.5を発表ã—ãŸã®ã¯ã€ãã®åˆé–“を縫ã£ã¦ã®ã“ã¨ã ã£ãŸã€‚ã‚‚ã¯ã‚„AI最å‰ç·šã®æˆ¦ã„ã¯ç ”究ãŒè¡Œã‚ã‚Œã‚‹å ´æ‰€ã®æ–‡åŒ–や風土ã«å½±éŸ¿ã‚’å—ã‘ã‚‹ã¨ã“ã‚ã¾ã§æ¥ã¦ã„る。 ãã—ã¦æ˜¥ç¯€ã‚‚ã‚ã‘ãŸä»Šé€±ã€ã•ã£ããAlibabaãŒã¨ã‚“ã§ã‚‚ãªã„トーã‚ングヘッドモデルを引ã£æã’ã¦ç™»å ´ã—ãŸã‹ã¨æ€ãˆã°ã€Microsoftã®ä¸å›½ãƒãƒ¼ãƒ ãŒã¨ã¦ã¤ã‚‚ãªã„LLMをリリース
Stable Diffusion 3 — Stability AI Japan
Stable Diffusion 3 ã®å…ˆè¡Œãƒ—レビューを発表ã—ã¾ã™ã€‚ã“ã®ãƒ¢ãƒ‡ãƒ«ã¯ã€è¤‡æ•°ä¸»é¡Œã®ãƒ—ãƒãƒ³ãƒ—トã€ç”»è³ªã€ã‚¹ãƒšãƒªãƒ³ã‚°èƒ½åŠ›ãªã©ã®ãƒ‘フォーマンスãŒå¤§å¹…ã«å‘上ã—ãŸã€é«˜æ€§èƒ½ãªç”»åƒç”Ÿæˆãƒ¢ãƒ‡ãƒ«ã§ã™ã€‚ ã“ã®ãƒ¢ãƒ‡ãƒ«ã¯ã¾ã 広ã利用å¯èƒ½ã§ã¯ã‚ã‚Šã¾ã›ã‚“ãŒã€æœ¬æ—¥ã€å…ˆè¡Œãƒ—レビューã®ã‚¦ã‚§ã‚¤ãƒ†ã‚£ãƒ³ã‚°ãƒªã‚¹ãƒˆã‚’開始ã—ã¾ã—ãŸã€‚ã“ã®ãƒ—レビュー段階ã¯ã€ã“ã‚Œã¾ã§ã®ãƒ¢ãƒ‡ãƒ«ã¨åŒæ§˜ã€ä¸€èˆ¬å…¬é–‹ã«å…ˆç«‹ã¡ã€æ€§èƒ½ã¨å®‰å…¨æ€§ã‚’å‘上ã•ã›ã‚‹ãŸã‚ã®çŸ¥è¦‹ã‚’åŽé›†ã™ã‚‹ãŸã‚ã«éžå¸¸ã«é‡è¦ã§ã™ã€‚ウェイティングリストã¸ã®ç™»éŒ²ã¯ã“ã¡ã‚‰ã‹ã‚‰ã€‚ Stable Diffusion 3ã®ä¸€é€£ã®ãƒ¢ãƒ‡ãƒ«ã¯ã€ç¾åœ¨800Mã‹ã‚‰8Bã®ãƒ‘ラメータをæä¾›ã—ã¦ã„ã¾ã™ã€‚ã“ã®ã‚¢ãƒ—ãƒãƒ¼ãƒã¯ã€ç§ãŸã¡ã®ã‚³ã‚¢ãƒãƒªãƒ¥ãƒ¼ã«æ²¿ã„ã€ã‚¢ã‚¯ã‚»ã‚¹ã‚’民主化ã™ã‚‹ã“ã¨ã‚’目的ã¨ã—ã¦ãŠã‚Šã€ãƒ¦ãƒ¼ã‚¶ãƒ¼ã®ã‚¯ãƒªã‚¨ã‚¤ãƒ†ã‚£ãƒ–ãªãƒ‹ãƒ¼ã‚ºã«æœ€é©ãªå½¢ã§å¿œãˆã‚‹ãŸã‚ã€æ‹¡å¼µæ€§ã¨å“質ã«é–¢ã™ã‚‹ã•ã¾ã–ã¾ãªã‚ªãƒ—ションをæä¾›ã—ã¾ã™ã€‚ Stable Diffus
Sora: Creating video from text
Sora Creating video from text Sora is an AI model that can create realistic and imaginative scenes from text instructions. Read technical report We’re teaching AI to understand and simulate the physical world in motion, with the goal of training models that help people solve problems that require real-world interaction. Introducing Sora, our text-to-video model. Sora can generate videos up to a mi
TikTokã¨ã®å¥‘ç´„ã«é–¢ã™ã‚‹ãŠçŸ¥ã‚‰ã› - UNIVERSAL MUSIC JAPAN
ユニãƒãƒ¼ã‚µãƒ«Â ミュージック グループ(UMG)ã®å…¬å¼ãƒ›ãƒ¼ãƒ ページã§ç™ºè¡¨ã•ã‚ŒãŸ(*注1)通りã€2024å¹´1月31日を期é™ã¨ã—ã¦ã„ãŸUMGã¨TikTokã®å¥‘ç´„ã«ã¤ã„ã¦ã¯ã€æ›´æ–°ã®åˆæ„ã«è‡³ã‚‹ã“ã¨ãŒã§ãã¾ã›ã‚“ã§ã—ãŸã€‚ 契約更新ã®äº¤æ¸‰éŽç¨‹ã‚’通ã˜ã¦ã€UMGã¯ä»¥ä¸‹3ã¤ã®é‡è¦ãªèª²é¡Œï¼š ・アーティストã¨ã‚½ãƒ³ã‚°ãƒ©ã‚¤ã‚¿ãƒ¼ã¸ã®é©åˆ‡ãªå ±é…¬ ・AIã®æœ‰å®³ãªå½±éŸ¿ã‹ã‚‰ã®äººé–“ã®ã‚¢ãƒ¼ãƒ†ã‚£ã‚¹ãƒˆã‚’守るã“㨠・TikTokユーザーã®ã‚ªãƒ³ãƒ©ã‚¤ãƒ³ä¸Šã®å®‰å…¨æ€§ ã«å¯¾å‡¦ã™ã‚‹ã‚ˆã†TikTokã«æ案ã—ã¦ã¾ã„ã‚Šã¾ã—ãŸãŒã€TikTokã¯ã“ã‚Œã«ç„¡é–¢å¿ƒã€æˆ–ã„ã¯å¨åœ§çš„ãªæ…‹åº¦ã§å¿œã˜ã¾ã—ãŸã€‚ UMGã¯ã“ã†ã—ãŸTikTokã®å§¿å‹¢ãŒå¼•ãèµ·ã“ã™å•é¡Œã‚’ã€ã“れ以上看éŽã™ã‚‹ã‚ã‘ã«ã¯ã„ã‹ãªã„ã¨è€ƒãˆã¦ã„ã¾ã™ã€‚ 契約ãŒåˆæ„ã«è‡³ã‚‰ãªã‹ã£ãŸã“ã¨ã«ä¼´ã„ã€ãƒ•ã‚¡ãƒ³ã®çš†æ§˜ã«ã¯TikTok上ã§UMGã®ç®¡ç†æ¥½æ›²(*注2)を利用ã—ãŸå‹•ç”»ã®ä½œæˆã‚„視è´ã«å½±éŸ¿ãŒå‡ºã‚‹å½¢ã¨ãªã‚Šã€ã”迷惑・ã”ä¸ä¾¿ã‚’ãŠã‹ã‘ã—ã¦
Gemini 㧠Bard ãŒã•ã‚‰ã«ä¾¿åˆ©ã«
本日ã€Google ã¯ã€ãƒžãƒ«ãƒãƒ¢ãƒ¼ãƒ€ãƒ«ã§ã®é«˜åº¦ãªæŽ¨è«–性能を備ãˆãŸé«˜æ€§èƒ½ AI モデル Gemini を発表ã—ã¾ã—ãŸã€‚柔軟性をé‡è¦–ã—ã¦è¨è¨ˆã•ã‚ŒãŸ Gemini ã¯ã€Ultraã€Proã€Nano ã® 3 サイズã«æœ€é©åŒ–ã•ã‚Œã¦ã„ã‚‹ãŸã‚ã€ãƒ‡ãƒ¼ã‚¿ã‚»ãƒ³ã‚¿ãƒ¼ã‹ã‚‰ãƒ¢ãƒã‚¤ãƒ« デãƒã‚¤ã‚¹ã¾ã§ã‚らゆる環境ã§å‹•ä½œã—ã¾ã™ã€‚ Gemini ã¯ã€Bard ã«ã‚‚アップデートã•ã‚Œã€ä»Šå¾Œã€€2 段階ã§æä¾›ã•ã‚Œã¾ã™ã€‚ã¾ãšã€æœ¬æ—¥ã‚ˆã‚Šã€Bardã§ã¯ç‰¹åˆ¥ã«èª¿æ•´ã•ã‚ŒãŸ Gemini Pro ãŒä½¿ã‚ã‚Œã¾ã™ã€‚（英語版ã®ã¿ï¼‰ãã—ã¦ã€æ¥å¹´åˆã‚ã«ã¯ã€Gemini Ultra ã§å‹•ä½œã™ã‚‹ã‚¢ãƒ‰ãƒãƒ³ã‚¹ モデル㮠Bard Advanced を公開ã—ã¾ã™ã€‚ Bard ã§Gemini Pro を体験一般公開ã«å…ˆç«‹ã¡ã€ç§ãŸã¡ã¯å¤šæ•°ã®æ¥ç•Œæ¨™æº–ベンãƒãƒžãƒ¼ã‚¯ã§ Gemini Pro をテストã—ã¾ã—ãŸã€‚ã“れらã®ãƒ™ãƒ³ãƒãƒžãƒ¼ã‚¯ã§ã¯ã€Gemini Pro ã¯å¤§è¦æ¨¡ãª AI
リリースã€éšœå®³æƒ…å ±ãªã©ã®ã‚µãƒ¼ãƒ“スã®ãŠçŸ¥ã‚‰ã›
最新ã®äººæ°—エントリーã®é…ä¿¡
j次ã®ãƒ–ックマーク
kå‰ã®ãƒ–ックマーク
lã‚ã¨ã§èªã‚€
eコメント一覧を開ã
oページを開ã
Training Large Language Models to Reason in a Continuous Latent Space
Google NotebookLM | Note Taking & Research Assistant Powered by AI
(PDF)文化審è°ä¼š 著作権分科会 法制度å°å§”員会「AIã¨è‘—作権ã«é–¢ã™ã‚‹è€ƒãˆæ–¹ã«ã¤ã„ã¦ã€|文化åºè‘—作権課
https://www.bunka.go.jp/seisaku/bunkashingikai/chosakuken/hoseido/r05_07/pdf/94024201_01.pdf
Home \ Anthropic
{{#tags}}- {{label}}
{{/tags}}