01News
The 2026 China Internet Conference will open in Beijing on July 8 and last for three days. The conference features a rich agenda of forums focusing on core areas such as artificial intelligence, computing power networks, and data security. Emerging topics such as the Agent Internet, Token economy, and Agent security have become focal points, aiming to explore new pathways for future industrial innovation and digital economic development.
Source links can be verified on the original siteRead original ↗02News
Google's 2025 Environmental Report shows that although the expansion of AI infrastructure caused the company's electricity consumption to surge 37% year-on-year, hitting a record high, operational carbon emissions fell 2% year-on-year. This is mainly due to large-scale procurement of renewable energy, improved data center efficiency, and optimized full-stack AI computing performance. Google has achieved 100% renewable energy electricity matching for nine consecutive years.
Source links can be verified on the original siteRead original ↗03News
Meta recently released social AI app Pocket, featuring the concept of "Vibe Coding", where users can generate mini-programs or mini-games using natural language without code, such as generating a brush from flower photos, generating a retro game console, etc. Generated content can be saved in "playlists" to share, or explore others' creations.
Source links can be verified on the original siteRead original ↗04News
Wealthy US families are increasingly sending their kids to AI schools like Alpha School, which combines two hours of AI tutoring with project-based workshops for up to $75,000 a year in tuition. The trend highlights a growing education gap in the AI era, where traditional schools are struggling to adopt the technology, which may do more harm than good if used without the right skills.
Source links can be verified on the original siteRead original ↗05News
Mistral founder Arthur Mensch warns companies against relying on closed AI models. He claims AI labs are storing more and more customer data and have, in some cases, used it to go after their own customers as competitors. The concern is valid, but Mistral can't really compete with frontier models from OpenAI or Anthropic on performance and is betting heavily on EU sovereignty as its strategic edge.
Source links can be verified on the original siteRead original ↗06News
Students need to use 10 questions they designed themselves to 'stump' three AI models; the more questions the AI answers wrong and the more models it stumps, the stronger the question-setting student's score is.
Source links can be verified on the original siteRead original ↗07News
Data shows that AI-native startups have teams 25% smaller, with the share of engineers about 13% higher, while the proportions of junior staff and management are about 15% lower respectively compared with non-AI-native companies.
Source links can be verified on the original siteRead original ↗08News
AI search agents rarely fail at multi-step research because of the search itself. Their real problem is not asking the user for clarification when queries are ambiguous. A new benchmark called DiscoBench shows that models searching repeatedly instead of asking follow-up questions actually perform worse, at 51.9 percent, than those that just guess. Even the best model only hits 43 percent overall accuracy. When ambiguity is removed from the queries, accuracy jumps by up to 40 points.
Source links can be verified on the original siteRead original ↗09News
LlamaIndex has published legal-kb, a public reference application on GitHub. It is described as a knowledge base for legal documents, powered by LlamaIndex Index v2 (the LlamaParse Platform). The project demonstrates a pattern the team calls a Retrieval Harness for agentic retrieval.
Source links can be verified on the original siteRead original ↗10News
Ubuntu is actively embracing AI, while Fedora community's 'AI desktop' plan is suspended due to strong opposition. The plan aimed to lower the barrier to AI programming, but the community considered the current proposal process immature. This reflects differing attitudes toward AI applications within the Linux ecosystem.
Source links can be verified on the original siteRead original ↗11News
A request for proposal released by Anthropic shows that, after setting up an Australian office earlier this year, the company aims to put at least 1 gigawatt of computing power into use by the end of next year. (Australian Financial Review)
Source links can be verified on the original siteRead original ↗12News
Don't panic! You will be automatically returned to the ITHome homepage in 5 seconds, please wait...
Source links can be verified on the original siteRead original ↗13News
LG Display, through comprehensive application of AI-driven virtual design verification technology, optimizes design schemes without making physical samples, greatly shortening the display panel prototype development cycle from three months. At the same time, its AI production system compresses quality anomaly handling time from three weeks to two days, and with higher yield rates, the company can save over 200 billion Korean won in costs per year.
Source links can be verified on the original siteRead original ↗14News
Japanese industrial robot giants Yaskawa Electric and Fanuc are actively embracing artificial intelligence, launching AI robots that can autonomously perceive and adjust movements, and collaborating with tech giants such as Nvidia and Google. Facing declining market share, Japanese companies attempt to seek breakthroughs in the new 'embodied intelligence' track through AI and open collaboration to regain their former global market dominance.
Source links can be verified on the original siteRead original ↗15News
Reuters reports that the Trump administration denied discussions about taking a stake in AI giant Anthropic. This clarification stems from speculation regarding similar negotiations with OpenAI. The US government is strengthening regulatory scrutiny of cutting-edge AI models to prevent misuse risks, while also exploring ways to allow the public to share in the industry's dividends.
Source links can be verified on the original siteRead original ↗16News
Summary I found an aggregate pattern in Codex token_count metadata: gpt-5.5 responses disproportionately land at exactly reasoning_output_tokens = 516, with additional fixed-boundary spikes around 1034 and 1552. This appears model-specif...
Source links can be verified on the original siteRead original ↗17News
I wrote about the sqlite-utils 4.0rc1 release a couple of weeks ago. Since we only have Claude Fable on our Max subscriptions for a few more days, I decided to …
Source links can be verified on the original siteRead original ↗18News
Microsoft, Google, Amazon and other giants plan to invest trillions of dollars in building AI infrastructure, but the 'indirect water consumption' of data centers and their supporting power generation far exceeds disclosed data, and many are located in water-stressed regions, which may trigger regional water competition in the future.
Source links can be verified on the original siteRead original ↗19News
Toyota plans to build a full-chain integrated system covering planning to sales by 2028, using AI to streamline 45,000 departmental terms into 5,000 sets of common standards, thereby cutting 30% of intermediate processes and improving supply chain efficiency.
Source links can be verified on the original siteRead original ↗20News
Junyang Lin was the technical lead of Alibaba’s Qwen project. He announced he was stepping down on March 3, 2026. He now lists himself as an independent researcher on his personal site.
Source links can be verified on the original siteRead original ↗