rui 的 blog

菜单

Administrator
发布于 2026-05-13 / 3 阅读
0
0

AI 每日资讯 - 2026-05-13

发布日期：2026-05-13

收录条目：20

1. Sam Altman was winning on the stand, but it might not be enough

摘要：After two weeks of hearing from assorted witnesses that he was a lying snake, the jury finally heard from the lying snake himself: Sam Altman. At the end of the testimony, his lawyer William Savitt asked him how it felt

2. Build a Hybrid-Memory Autonomous Agent with Modular Architecture and Tool Dispatch Using OpenAI

摘要：In this tutorial, we begin by exploring the architecture behind a hybrid-memory autonomous agent. This system combines semantic vector search, keyword-based retrieval, and a modular tool-dispatching loop to create an age

3. Meet AntAngelMed: A 103B-Parameter Open-Source Medical Language Model Built on a 1/32 Activation-Ratio MoE Architecture

摘要：MedAIBase has released AntAngelMed, a 103B-parameter open-source medical language model that uses a 1/32 activation-ratio Mixture-of-Experts (MoE) architecture to activate only 6.1B parameters at inference time, matching

4. Meta won’t let you block its AI account on Threads

摘要：Meta announced on Tuesday that it's testing a Threads feature that lets users tag a Meta AI account to get answers to questions or context about a conversation on the platform. If you've spent any time looking at replies

5. Sam Altman says Elon Musk’s mind games were damaging OpenAI

摘要：OpenAI CEO Sam Altman says Elon Musk did "huge damage" to the culture of the AI startup. During testimony as part of Musk's lawsuit against OpenAI, Altman said Musk required OpenAI president Greg Brockman and former chie

6. The 9 biggest new features in Android 17

摘要：Would it shock you to hear that Android 17 is filled with new AI-enabled features, like improved dictation and vibe-coded widgets? Fortunately, that's not all. The platform is getting non-AI updates too, from an emoji ov

7. Gemini’s latest updates are all about controlling your phone

摘要：It is, once again, Gemini season. Google is announcing a host of new Gemini features during its pre-I/O Android showcase, many of which aim to help use your phone for you. You'll find Gemini in more places, like Chrome o

8. Parents say ChatGPT got their son killed with bad advice on party drugs

摘要：The family of a 19-year-old college student is suing OpenAI over claims that his conversations with ChatGPT led to an accidental overdose. In the lawsuit filed on Tuesday, Sam Nelson's parents allege ChatGPT "encouraged"

9. Sam Altman takes the stand in trial against Elon Musk

摘要：OpenAI CEO Sam Altman has begun his testimony against Elon Musk in a high-profile jury trial in a California federal courtroom. Altman, alongside OpenAI president Greg Brockman, is a primary defendant in the trial brough

10. George Clooney, Tom Hanks, and Meryl Streep back new ‘Human Consent Standard’ for AI licensing

摘要：Hollywood actors and producers are standing behind a new AI licensing standard that will tell AI systems whether they'll need to pay to use a person's likeness, creative work, characters, and designs. With the Human Cons

11. Rivian’s AI-powered voice assistant is ready to roll

摘要：Rivian's AI-powered voice assistant is rolling out today to the company's vehicle fleet. The assistant will be available through a software update to all compatible Rivian Gen 1 and Gen 2 vehicle owners who subscribe to

12. How finance teams use Codex

摘要：See how finance teams can use Codex to build MBRs, reporting packs, variance bridges, model checks, and planning scenarios from real work inputs.

13. Tilde Research Introduces Aurora: A Leverage-Aware Optimizer That Fixes a Hidden Neuron Death Problem in Muon

摘要：Researchers at Tilde Research have released Aurora, a new optimizer for training neural networks that addresses a structural flaw in the widely-used Muon optimizer. The flaw quietly kills off a significant fraction of ML

14. A Coding Implementation to Portfolio Optimization with skfolio for Building Testing, Tuning, and Comparing Modern Investment Strategies

摘要：In this tutorial, we explore skfolio, a scikit-learn compatible portfolio optimization library that helps us build, compare, and evaluate different investment strategies in a structured Python workflow. We start by loadi

15. OpenAI Introduces Daybreak: A Cybersecurity Initiative That Puts Codex Security at the Center of Vulnerability Detection and Patch Validation

摘要：OpenAI on just launched Daybreak, a cybersecurity initiative that combines the company’s frontier AI models with Codex Security, its coding-focused agentic system, and a broad network of security partners. The program is

16. Where Reliability Lives in Vision-Language Models: A Mechanistic Study of Attention, Hidden States, and Causal Circuits

摘要：arXiv:2605.08200v1 Announce Type: new Abstract: A pervasive intuition holds that vision-language models (VLMs) are most trustworthy when their attention maps look sharp: concentrated attention on the queried region shoul

17. Spatial Priming Outperforms Semantic Prompting: A Grid-Based Approach to Improving LLM Accuracy on Chart Data Extraction

摘要：arXiv:2605.08220v1 Announce Type: new Abstract: The automated extraction of data from scientific charts is a critical task for large-scale literature analysis. While multimodal Large Language Models (LLMs) show promise,

18. Auto-Rubric as Reward: From Implicit Preferences to Explicit Multimodal Generative Criteria

摘要：arXiv:2605.08354v1 Announce Type: new Abstract: Aligning multimodal generative models with human preferences demands reward signals that respect the compositional, multi-dimensional structure of human judgment. Prevailin

19. Embeddings for Preferences, Not Semantics

摘要：arXiv:2605.08360v1 Announce Type: new Abstract: Modern AI is opening the door to collective decision-making in which participants express their views as free-form text rather than voting on a fixed set of candidates. A n

20. On Distinguishing Capability Elicitation from Capability Creation in Post-Training: A Free-Energy Perspective

摘要：arXiv:2605.08368v1 Announce Type: new Abstract: Debates about large language model post-training often treat supervised fine-tuning (SFT) as imitation and reinforcement learning (RL) as discovery. But this distinction is

NoSuchUpload 复盘：为...

评论