Prediction, syntax and semantic grounding in the brain and large language models - Scientific Reports
Emerging from cutting-edge AI research, large language models (LLMs) such as GPT4o and Llama have revolutionized the way we interact with machines, surprising users with their human-like responses1,2,3,4. These models rely almost entirely on predictive processes to generate language5,6, enabling them to approach - or even surpass - the Turing-Test with unprecedented proficiency7,8. The human brain
How AI Impacts Skill Formation
How AI Impacts Skill Formation Judy Hanwen Shen∗ Alex Tamkin† February 3, 2026 Abstract AI assistance produces significant productivity gains across professional domains, particularly for novice workers. Yet how this assistance affects the development of skills required to effectively supervise AI remains unclear. Novice workers who rely heavily on AI to complete unfamiliar tasks may compromise th
Collaborative Text Editing with Eg-walker: Better, Faster, Smaller
Collaborative text editing algorithms allow several users to concurrently modify a text file, and automatically merge concurrent edits into a consistent state. Existing algorithms fall in two categories: Operational Transformation (OT) algorithms are slow to merge files that have diverged substantially due to offline editing; CRDTs are slow to load and consume a lot of memory. We introduce Eg-walk
エーザイ株式会社｜ヒューマン・ヘルスケアのエーザイ
2026年3月27日 研究開発 製品情報 「レンビマ®」（レンバチニブ）について、日本において、がん化学療法後に増悪した腎細胞がんを対象に「ウェリレグ®」（ベルズチファン）との併用療法に係る用法・用量を追加申請 2026年3月27日 研究開発 抗がん剤タレトレクチニブについて進行ROS1陽性 非小細胞肺がん治療に係る販売承認申請が欧州医薬品庁より受理 —本販売承認申請は標準スケジュールによるフル承認に向けた審査対象として受理 英国やカナダをはじめとするその他のエーザイのライセンス地域でも申請を予定 タレトレクチニブは、既に米国、中国、日本において、進行ROS1陽性非小細胞肺がんに係る適応で承認済— 2026年3月26日 企業情報 エーザイ株式会社 人事異動ならびに組織改編 2026年3月26日 企業情報 取締役候補者の選任に関するお知らせ 2026年3月25日 エコシステム 家族の会話から
Defeating Prompt Injections by Design
Large Language Models (LLMs) are increasingly deployed in agentic systems that interact with an untrusted environment. However, LLM agents are vulnerable to prompt injection attacks when handling untrusted data. In this paper we propose CaMeL, a robust defense that creates a protective system layer around the LLM, securing it even when underlying models are susceptible to attacks. To operate, CaMe
Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity
Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity We conduct a randomized controlled trial (RCT) to understand how early-2025 AI tools affect the productivity of experienced open-source developers working on their own repositories. Surprisingly, we find that when developers use AI tools, they take 19% longer than without—AI makes them slower. We view this resu
BioCLIP: A Vision Foundation Model for the Tree of Life
BioCLIP correctly labels this Clitocybe fragrans from the Fungi task. BioCLIP correctly labels this Jasmine leaf (Jasminum) from the Medicinal Leaf task. BioCLIP correctly labels this Rhizosolenia from the Plankton task. BioCLIP performs well on a variety of image sources, like this microscope image. CLIP mislabels this Russula ochroleuca as "the prince" (Agaricus augustus), which isn't even in th
Language Models Learn to Mislead Humans via RLHF
Jiaxin Wen1, Ruiqi Zhong2, Akbir Khan3, Ethan Perez3, Jacob Steinhardt2 Minlie Huang1, Samuel R. Bowman3434{}^{3~{}4}, He He4, Shi Feng4,5 1Tsinghua University 2Univeristy of California, Berkeley 3Anthropic 4New York University 5George Washington University Abstract Language models (LMs) can produce errors that are hard to detect for humans, especially when the task is complex. RLHF, the most pop
DexGen
Dexterity Gen: Foundation Controller for Unprecedented Dexterity Zhao-Heng Yin1,2, Changhao Wang2, Luis Pineda2, Francois Hogan2, Krishna Bodduluri2, Akash Sharma2, Patrick Lancaster2, Ishita Prasad2, Mrinal Kalakrishnan2, Jitendra Malik2, Mike Lambeta2, Tingfan Wu2, Pieter Abbeel1, and Mustafa Mukadam2. 1BAIR, Berkeley EECS and 2FAIR at Meta Achieving Unprecedented Dexterity We introduce Dexerity
60年前からのプログラミング和文論文を掘り起こした話
本記事の初版を公開したのは 2024 年 9 月 21 日だったのですが、その後 9 月 26 日に、ピックアップした論文を大幅に増やして更新しました。 また、最初はタイトルを「和文プログラミング論文を〜」としていたのですが、いわゆる「日本語プログラミング言語」の話にも見えてしまうことから、「プログラミング和文論文を〜」に置き換えました。 日本の人々が「プログラミング」にかかわるようになったのは、いつごろでしょうか? その前に、まずは世界のコンピュータとプログラミングの歴史を簡単に振り返ってみましょう。まずは OS から有名どころをたどると、 わかりやすく Windows 95 が 1995 年 [1] その前に広く使われた Windows 3.1 は 1992 年 [2] その下で動く MS-DOS は、バージョン 6 が 1993 年、バージョン 1 が 1981 年 [3] Linu
Globally representative evidence on the actual and perceived support for climate action - Nature Climate Change
The world’s climate is a global common good and protecting it requires the cooperative effort of individuals across the globe. Consequently, the ‘human factor’ is critical and renders the behavioural science perspective on climate change indispensable for effective climate action. Despite its importance, limited knowledge exists regarding the willingness of the global population to cooperate and a
リリース、障害情報などのサービスのお知らせ
最新の人気エントリーの配信
j次のブックマーク
k前のブックマーク
lあとで読む
eコメント一覧を開く
oページを開く
https://www.acm.org/publications/message-to-the-community
Anchor Sponsor Firms in Open Source Software Ecosystems
Robust-first computing: Demon Horde Sort (full version)
https://dl.acm.org/doi/10.1145/3593856.3595909
LLMmap: Fingerprinting For Large Language Models
https://scnps.co/papers/sp23_domclob.pdf
Bachelor’s Thesis
GitHub - papers-we-love/papers-we-love: Papers from the computer science community to read and discuss.
What Do Small Businesses Do?