Loading / 加载中
Thinkgap feed
Curated items are read from the processed items table and served as a bilingual feed.
6 items
A Telegram user posted that they saw something at the Zhangjiang Embodied Intelligence Conference. The message includes no further details about what was observed, such as companies, products, demos, or research. It provides no actionable or newsworthy information.
A neuroscience researcher described Neuralink as scientifically opportunistic and behind academic standards, favoring engineering over rigorous science. She detailed that their lab builds custom 22nm low-power chips for implantable neural recording, requiring up to 10 PB of storage to capture 30 kHz signals from hundreds of electrodes—raw data at 80 GB per probe per hour. The cost of maintaining PB-scale storage can reach millions of dollars over five years, forcing labs to buy only 1 PB at a time. She noted that Harvard and Google's 1 mm³ human brain dataset is 1.4 PB, while a whole mouse brain would be ~1 EB. The conversation also covered the shift from copper to optical interconnects, echoing NVIDIA's recent silicon photonics switch for AI data centers, as copper cannot sustain 1.6 Tb/s over rack-scale distances.
The post recalls a Jiqizhixin report on the Pangu NLP model after Yu Chengdong’s recent mention of it. Pangu was developed by Huawei Cloud and Cycle Intelligence, Yang Zhilin’s previous company, and the report already referred to the team as “NLP Moonshot”. It notes that before Pangu, Chinese NLP model competition was fierce, with Meituan, Alibaba, Sogou, and the GLM/CPM models under Wudao already active. The article includes a prediction by Tang Jie and Yang Zhilin about the coming AI era’s two features: a leap in AI production efficiency and exponential growth in AI application scenarios. The post suggests that those interested in the original Pangu team’s latest work could look at Kimi Moonshot products.
A Telegram post from channel @aigc1024 states that in the past, Leonardo da Vinci was knowledgeable in everything, and that this will also be true for humans in the future. The statement does not mention any specific AI technology, model, product, or development. It is a brief, philosophical observation about human potential.
The message contains only the text '4/8467)' with no further explanation, links, or media. No concrete subject, event, or factual information about AI news, models, products, papers, tools, or finance is present. The content is insufficient to derive any meaningful takeaway.
A SemiAnalysis article examines how Unitree Robotics achieves cost dominance not by squeezing suppliers but by adopting a more aggressive technical route: QDD (Quasi-Direct Drive) actuators instead of harmonic reducers. This choice trades early reliability risks (overheating, skepticism) for simpler planetary gearboxes, standard industrial components, and diverse suppliers. Iteration speed reaches weekly rather than quarterly cycles. About 200 Chinese humanoid robot companies are now switching to QDD, reinforcing the ecosystem and amplifying Unitree's advantage.