May 10 AI Briefing · Issue #279

2026-05-10 00:00

Author: RadarAI Editorial Editor: RadarAI Editorial Last updated: 2026-05-10 Review status: Editorial review pending Brief 速报官方 AI动态开源

The AI industry is rapidly shifting from model-centric competition to a race in systems engineering capability: Embodied intelligence relies on high-quality, closed-loop human behavioral data; multimodal reasoning focuses on 'visual primitives' to bridge the referential gap; and foundational advances—including sparse Transformers and AI-native knowledge graphs—are accelerating in parallel. Meanwhile, the OpenAI courtroom showdown and Michael Burry's bubble warning inject critical rationality into an overheated market [5][6].

Editorial standards and source policy: Editorial standards, Team. Content links to primary sources; see Methodology.

## 🔍 Core Insights The AI industry is rapidly shifting from **model-centric competition** to a race in **systems engineering capability**: **Embodied intelligence** depends on high-quality, closed-loop human behavioral data; **multimodal reasoning** centers on 'visual primitives' to resolve the referential gap; foundational breakthroughs—including **sparse Transformers** and **AI-native knowledge graphs**—are accelerating in tandem; while the **OpenAI courtroom showdown** and **Michael Burry's bubble warning** inject critical rational scrutiny into an overheated market [5][6]. ## 🚀 Key Updates - **'Paper Phone' Goes Viral with 100M+ Views: The Secret to AI-Driven Visual Storytelling Is 'Bloopers'** [0]: Creators confirm that 'imperfect,' human-centered narratives—combined with precise prompting—act as powerful amplifiers of emotional resonance. - **DeepSeek Releases Multimodal Technical Report with Peking & Tsinghua Universities** [7]: Proposes a referential mechanism grounded in 'points and bounding boxes' as visual primitives—directly tackling the 'referential gap' in complex visual reasoning. - **Zhejiang University & Shanghai AI Lab Launch SciGraph-SCP—the First AI-Native Scientific Knowledge Graph** [15]: Integrates 370M+ entities across 68 subgraphs, empowered by the standardized SCP protocol to fuel scientific agents. - **StepAudio 2.5 TTS by Jieyue (Step) Tops Artificial Analysis' Blind Test in China** [13]: Delivers natural listening experiences matching international top-tier standards—ranked among the world's top three. - **AutoNavi's ABot-NeoVerse Wins ICRA 2026 AGIBot Global Challenge** [20]: Its in-house world model validates the generational advantage of full-stack embodied AI in physical consistency and synthetic data generation. - **Xiaomi Open-Sources SVOR—A Video Object Removal Framework** [12]: Three core innovations effectively address shadow residue and motion jitter, advancing industrial-grade video editing. - **Perplexity Publishes Internal Agent Skills Design Guidelines** [2]: Emphasizes that Skill design is fundamentally about constructing context—not writing code—and defines an iterative, maintainable paradigm for agent capability development. - **U.S. Researcher Visits Over Ten Chinese Labs: 'Lack of Compute' Is Consensus—but Engineering Rigor and Deployment Mindset Are Rapidly Closing the Model Gap** [16]: ByteDance is widely viewed with apprehension; DeepSeek earns 'profound respect'—highlighting differentiated ecosystem competitiveness. ## 🔗 Sources [0] An AI Short Film With 100M+ Views—Its Iconic Scene Is a 'Blooper': Interview with the Creators of 'Paper Phone' — https://www.bestblogs.dev/article/dc6a8687?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [2] Perplexity's Internal Playbook for Designing, Iterating, and Maintaining Agent Skills — https://www.bestblogs.dev/status/2053098123074105705?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [5] The Big Short: Today's AI Boom Resembles the Internet Bubble 'Months Before It Burst' — https://www.bestblogs.dev/article/05ca1ade?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [6] All Gloves Off! OpenAI Trial Turns Ugly: Musk Called 'Impatient and AI-Ignorant'; Altman Labeled 'Hypocritical and Morally Flawed' — https://www.bestblogs.dev/article/4c214882?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item

The AI industry is rapidly shifting from model-centric competition to a race in systems engineering capability: Embodied intelligence depends on high-quality, closed-loop human behavioral data; multimodal reasoning centers on 'visual primitives' to resolve the referential gap; foundational breakthroughs—including sparse Transformers and AI-native knowledge graphs—are accelerating in tandem; while the OpenAI courtroom showdown and Michael Burry's bubble warning inject critical rational scrutiny into an overheated market [5][6].

🚀 Key Updates

'Paper Phone' Goes Viral with 100M+ Views: The Secret to AI-Driven Visual Storytelling Is 'Bloopers' [0]: Creators confirm that 'imperfect,' human-centered narratives—combined with precise prompting—act as powerful amplifiers of emotional resonance.
DeepSeek Releases Multimodal Technical Report with Peking & Tsinghua Universities [7]: Proposes a referential mechanism grounded in 'points and bounding boxes' as visual primitives—directly tackling the 'referential gap' in complex visual reasoning.
Zhejiang University & Shanghai AI Lab Launch SciGraph-SCP—the First AI-Native Scientific Knowledge Graph [15]: Integrates 370M+ entities across 68 subgraphs, empowered by the standardized SCP protocol to fuel scientific agents.
StepAudio 2.5 TTS by Jieyue (Step) Tops Artificial Analysis' Blind Test in China [13]: Delivers natural listening experiences matching international top-tier standards—ranked among the world's top three.
AutoNavi's ABot-NeoVerse Wins ICRA 2026 AGIBot Global Challenge [20]: Its in-house world model validates the generational advantage of full-stack embodied AI in physical consistency and synthetic data generation.
Xiaomi Open-Sources SVOR—A Video Object Removal Framework [12]: Three core innovations effectively address shadow residue and motion jitter, advancing industrial-grade video editing.
Perplexity Publishes Internal Agent Skills Design Guidelines [2]: Emphasizes that Skill design is fundamentally about constructing context—not writing code—and defines an iterative, maintainable paradigm for agent capability development.
U.S. Researcher Visits Over Ten Chinese Labs: 'Lack of Compute' Is Consensus—but Engineering Rigor and Deployment Mindset Are Rapidly Closing the Model Gap [16]: ByteDance is widely viewed with apprehension; DeepSeek earns 'profound respect'—highlighting differentiated ecosystem competitiveness.

🔗 Sources

[0] An AI Short Film With 100M+ Views—Its Iconic Scene Is a 'Blooper': Interview with the Creators of 'Paper Phone' — https://www.bestblogs.dev/article/dc6a8687?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item
[2] Perplexity's Internal Playbook for Designing, Iterating, and Maintaining Agent Skills — https://www.bestblogs.dev/status/2053098123074105705?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item
[5] The Big Short: Today's AI Boom Resembles the Internet Bubble 'Months Before It Burst' — https://www.bestblogs.dev/article/05ca1ade?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item
[6] All Gloves Off! OpenAI Trial Turns Ugly: Musk Called 'Impatient and AI-Ignorant'; Altman Labeled 'Hypocritical and Morally Flawed' — https://www.bestblogs.dev/article/4c214882?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item

← Back to Updates