Every story we have kept, newest first.
Looking for the daily editions? → Past editions
Page 341 of 1375
NEC has launched a new marketing service that leverages Anthropic's Claude to fully automate product planning and promotional plan creation based on consumer purchase data. The company aims to achieve 10 billion yen in cumulative sales within three years.
ITmedia AI+ (日本) · 83d ago
The marketing logic behind celebrity athlete endorsements is shifting from emotional resonance to high-risk, high-reward 'abstract' engagement. Brands are re-evaluating the ROI of traditional star-power campaigns.
钛媒体 · 83d ago
Tsinghua University's elite math program is struggling with students who gain admission through rote memorization and specialized test prep rather than genuine mathematical talent. This has led to high failure rates in advanced coursework and calls for the dismissal of unqualified students.
虎嗅 · 83d ago
India's software outsourcing sector, fueled by 1.5 million annual computer science graduates, faces disruption from AI automation. The traditional model of low-cost labor is being challenged by AI's ability to perform coding tasks more efficiently.
cnBeta (Full RSS) · 83d ago
South Korean antitrust regulators have fined eight local SolidWorks software distributors 2.3 billion KRW for colluding to manipulate software prices.
36氪 · 83d ago
The article analyzes the intensifying competition in the humanoid robot market between Unitree and UBTECH. It highlights the shifting landscape of embodied AI as these companies vie for industry dominance.
钛媒体 · 83d ago
Li Kaifu's 01.AI is shifting focus from general model training to B2B enterprise decision-making AI. The company aims to help CEOs navigate complex data to make better decisions, targeting 1.5 billion RMB in orders by 2026.
虎嗅 · 83d ago
TPG and Blackstone are planning to sell the surgical business unit of Hologic, targeting a valuation of over $4 billion. The move aims to repay debt and recover capital from their recent privatization of the company.
36氪 · 83d ago
As AI labs like DeepMind pursue AGI, philosophers are increasingly vital in addressing ethical alignment, bias, and the long-term impact of superintelligence. The article explores the tension between technical optimization and human-centric value systems.
虎嗅 · 83d ago
X is developing a feature that notifies users when a post they have interacted with—via likes, replies, or reposts—receives a Community Note. This aims to increase transparency regarding misinformation across the platform.
Digital Trends · 83d ago
Google Search reached an all-time high in usage during the Argentina vs. Egypt World Cup match.
IT之家 · 83d ago
Genshin Impact 7.0 introduces updated hardware requirements, specifically flagging potential compatibility risks for devices using certain Mali and Imagination GPU architectures. Most configuration requirements remain unchanged for major platforms.
IT之家 · 83d ago
The article discusses the tension between the rapid speed of AI and the need for thoughtful, deliberate government decision-making. It warns that prioritizing speed over reflection can lead to systemic errors and accountability gaps.
虎嗅 · 83d ago
Kongshanci is developing a brain-computer interface (BCI) platform aimed at treating mental and cognitive disorders. By combining signal analysis with定向 (targeted) physical energy intervention, they aim to restore stability to disordered brain states.
虎嗅 · 83d ago
LingBot-World 2.0 introduces a new paradigm for AI-generated environments that feature infinite duration and randomized, evolving elements. This update aims to create more immersive and 'living' digital spaces for users.
Ifanr (爱范儿) · 83d ago
Google confirmed that search traffic reached an all-time high during the Argentina vs. Egypt World Cup match.
cnBeta (Full RSS) · 83d ago
Google is mandating the migration to Manifest v3 (MV3) for all Chrome extensions, with MV2 support ending on August 31. This transition aims to improve browser security and resource efficiency but impacts the efficacy of ad-blocking tools.
cnBeta (Full RSS) · 83d ago
Zhijian Dynamics has successfully delivered its first 100 i7 Pro robots for CNC machining, achieving rapid deployment in under 10 days and establishing a new benchmark for industrial embodied AI.
雷峰网 · 83d ago
Researchers introduced a ReAct-style agentic framework that integrates LLM reasoning with SageMath for verifiable mathematical computation. The study demonstrates significant performance gains across various models, narrowing the gap between open-weight and closed-source systems.
ArXiv AI · 83d ago
This research demonstrates that the orchestration layer, rather than just the underlying model, is the primary lever for optimizing enterprise AI token economics. By using a specialized harness, tasks saw a 41% reduction in cost and 44% faster execution while maintaining quality.
ArXiv AI · 83d ago
This research proposes a new paradigm for evaluating AI models that have surpassed human capability by using relative measurement instead of static benchmarks. The framework utilizes model-generated challenges to create an adversarial psychometric rating system that scales alongside agent capabilities.
ArXiv AI · 83d ago
Researchers introduced the HALE framework, which integrates LLMs into agent-based modeling to simulate human decision-making dynamically. This approach overcomes the limitations of static priors in traditional simulations, demonstrated through a COVID-19 case study in Salt Lake County.
ArXiv AI · 83d ago
Researchers identified three key principles of human social norms—predictability, value alignment, and advantage awareness—to improve AI interaction. Integrating these norms into LLMs resulted in a fourfold performance increase in dynamic coordination tasks compared to baseline models.
ArXiv AI · 83d ago
Researchers introduced the Large Behavioral Model (LBM), which uses transaction history and retrieval-augmented generation to simulate customer decision-making. It outperforms general-purpose LLMs in retail-specific tasks like basket completion and promotion response.
ArXiv AI · 83d ago
研究人員發現目標導向的世界模型常出現「指令洩漏」現象，模型會直接從指令中提取答案而非真正理解空間關係。透過將目標從動態預測中分離並監督讀取路徑，可恢復模型真正的感知能力。
ArXiv AI · 83d ago
AgentLens is a new open-source benchmark that evaluates coding agents based on their entire interaction trajectory rather than just binary pass/fail results. It combines formal verification with LLM-generated reviews to provide actionable insights into agent behavior and performance.
ArXiv AI · 83d ago
Researchers developed an Explorer-Definer pipeline and Reflective Orchestrator to improve ARC-AGI-1 performance using DeepSeek V3.2. This approach achieves significant gains in abstract reasoning without benchmark-specific training or heavy test-time compute.
ArXiv AI · 83d ago
A recent scientific study suggests that the appeal of attractive characters in games and media is universal across genders, challenging the DEI-driven narrative that such designs are exclusively for male gratification.
cnBeta (Full RSS) · 83d ago
The LaST-HD research, led by Peking University and Zhijian Dynamics, proposes using human hand data to teach robots underlying physical laws rather than just mimicking motion, improving generalization.
雷峰网 · 83d ago
Tianli Qiming's AI-driven education solution has been recognized by UN agencies. It is now part of the AI for Good initiative, focusing on personalized learning paths.
量子位 · 83d ago