News came on July 4 that the US government will restrict Chinese companies from using cloud computing services from US manufacturers, and plug the loopholes that Chinese artificial intelligence companies may bypass the current chip export control rules by using cloud services. The bad news came, and Chinese technology companies were not afraid of danger. In Shanghai, an unprecedented artificial intelligence conference showed the vitality of China's artificial intelligence to the outside world: more than 30 large-scale models were unveiled, 177,000 offline audiences, 210 exhibitors, Intended purchase amount of 11 billion yuan. A domestic venture capital institution shouted directly at the scene: In 2024, China will have a multilingual general-purpose model comparable to GPT-4. Musk said that once China decides to do something, it will be able to do it well, and China will have strong artificial intelligence capabilities. The founder of midjourney also confessed that the name of the midjourney project was inspired by Zhuangzi, a thinker more than 2,000 years ago. Brace yourself, AIGC Entrepreneurs!
Image source: Generated by Unbounded AI tool
Financing
Beilian Zhuguan completed the angel round financing of 50 million yuan, and its business covers the AI computing power market
According to ChinaVenture.com, recently, Zhejiang Beilian Zhuguan, a cloud resource management service provider, announced the completion of an angel round of financing of 50 million yuan, invested by Yuanjing Capital, Sequoia China Seed Fund and Zhouxuan Equity.
Shanda Network CEO Chen Tianqiao invested another 1 billion yuan to support AI brain science
According to a report by the Financial Associated Press on July 6, Tianqiao Brain Science Research Institute (TCCI) announced that it will invest another 1 billion yuan to support AI brain development at a forum themed on "Brain-Computer Intelligence Digital Life" at the Shanghai World Artificial Intelligence Conference (WAIC). Science, including the establishment of MindX laboratory, the construction of a series of cutting-edge laboratories such as AI + sleep dreamland, AI + anti-aging, each with a start-up fund of 50 million yuan, and will continue to invest.
AI start-up company "Fourth Paradigm" has been approved by the China Securities Regulatory Commission for IPO in Hong Kong
According to an announcement on the official website of the China Securities Regulatory Commission, Beijing Fourth Paradigm Technology Co., Ltd., a domestic AI start-up company, has completed the relevant procedural work for an IPO on the Hong Kong Stock Exchange and has been approved by the China Securities Regulatory Commission. It is expected to conduct a listing hearing in the near future.
big model
Data: The number of domestic large-scale models has reached 93, and the proportion of Beijing has dropped to 45% for the first time
On July 4th, according to the statistics of the WeChat public account "Towards the Future", the number of domestic large-scale models has reached 93, which is about to exceed 100, and the proportion of Beijing has dropped to 45% for the first time.
More than 30 large models will be unveiled at the World Artificial Intelligence Conference
According to Times Finance, at the 6th World Artificial Intelligence Conference that just passed, more than 30 large-scale models were unveiled, and more than 20 large-scale model forums were overwhelmed, and the relevant booths were crowded with visitors without exception. Among the more than 30 large models exhibited at the venue, there are many GM models from major manufacturers such as Baidu Wenxin Yiyan, Alibaba Tongyi Qianwen, Huawei Pangu Large Model, HKUST Xunfei Xinghuo Cognition, and Shangtang Ririxin Sensenova. large model. There are also vertical large-scale models launched by small and medium-sized manufacturers such as Lanzhou Technology, Lianyuan Technology, Mobvoi, Xinghuan Technology, Midu, and Daguan Data for segmented scenarios.
Gao Wen, academician of the Chinese Academy of Engineering: will build a 200 billion parameter natural language large model base, and its performance will be benchmarked against ChatGPT
According to a report by IT House on July 6, Gao Wen, academician of the Chinese Academy of Engineering and director of Pengcheng Laboratory, said at the WAIC2023 Shengteng Artificial Intelligence Industry Summit Forum that Pengcheng Laboratory has launched a large-scale brain model project, aiming to build the first domestic Completely independent innovation, open source and open natural language pre-training large model base, the parameter level reaches 200 billion, and the performance is benchmarked against ChatGPT.
OpenAI fully opens the GPT-4 API and discards some old Completions API models
Babbitt News, on July 7, OpenAI announced on its official website that the GPT-4 API has been fully opened for use, and all paid API users can directly access GPT-4 in 8K context without any waiting. OpenAI is expected to open the GPT-4 API to new developers by the end of July. OpenAI also introduced GPT-3.5 Turbo, DALL·E and Whisper API. Fine-tuned versions of the safe and reliable GPT-4 and GPT-3.5 Turbo will also be available, expected later this year.
Shanghai Artificial Intelligence Laboratory released a general large-scale model system for scholars
According to the "Kechuangban Daily" report on July 6, the Shanghai Artificial Intelligence Laboratory launched a general large-scale model system for scholars at the 2023 World Artificial Intelligence Conference Science Frontiers Plenary Session, including three bases: multimodal, Puyu, and Tianji. At the same time, it launched the first full-chain open source system for the development and application of large models.
Among them, the multi-modal large model has 20 billion parameters, supports 3.5 million semantic tags, and is world-leading in 80+ tasks; the Pu language large model is the first large model officially released in China with 100 billion parameters supporting multiple languages; The model is the world's first city-level NeRF model.
Huawei Cloud Pangu Large Model 3.0 is officially released
According to a report by the Financial Associated Press on July 7, at the Huawei Developer Conference 2023 (Cloud) held this afternoon, Huawei Executive Director and Huawei Cloud CEO Zhang Pingan announced that Huawei Cloud Pangu Model 3.0 was officially released.
Zhang Pingan said that the Pangu Large Model 3.0 is divided into a three-layer structure of L0 basic model, L1 industry model, and L2 scene model, which will reshape thousands of industries. According to reports, this is a large model completely oriented to the industry. Among them, the Pangu L0 layer basic large model includes a natural language large model with hundreds of billions of parameters, Pangu multi-modal large model, visual large model, prediction large model, scientific computing large model, etc. The L1 layer is a large industry model, including government affairs , finance, manufacturing, mining, etc. The L2 layer is the scene model.
Tianlang releases three products including a general language model and a semantic search engine
According to the "Kechuangban Daily" report on July 8, at the World Artificial Intelligence Conference, Tianrang released the large-scale product ecology "Tiankang Xiaobai", which includes three products, namely a large language model, a semantic search engine and a development platform. It is understood that the Tianjie Xiaobai universal language large model provides ChatGPT-like capabilities through deep learning training with 186 billion parameters, and has the capabilities of multi-language dialogue interaction, knowledge question and answer, and logical reasoning. The semantic search engine enhances the generation ability of large models based on accurate facts, so that it can output based on accurate facts and avoid misleading information that large models may produce.
China Mobile released two artificial intelligence industry models
According to a report by the Financial Associated Press on July 8, China Mobile officially released the Jiutian Haisuan government affairs model and the Jiutian customer service model. It is reported that the Jiutian Haisuan government affairs model is an industry-oriented model for government affairs created by China Mobile based on the rich experience in digital government construction accumulated in recent years. Relying on the Jiutian Haisuan government affairs model, the government service system will have a powerful government affairs Item understanding ability, multi-dimensional information association ability, and multi-interaction ability for complex issues and complex processes; Jiutian·Customer Service Large Model can analyze user questions and provide answers based on natural language descriptions provided by users, and can also cooperate with human customer service , analyze the semantics and context of historical communication content, summarize and summarize the key points and key information of the dialogue, and prompt and reply suggestions for the artificial customer service agent according to the dialogue content, so as to reduce the response time and form a three-way communication field of "big model-manual agent-user" , greatly improving the work efficiency of traditional manual customer service.
China Telecom announced the TeleChat large model, claiming that it will establish a "tens of billions AI service provider"
According to a report from IT House on July 3, at the Artificial Intelligence Summit Forum of the 2023 Global Digital Economy Conference held yesterday, Zhang Xin, deputy general manager of China Telecom Group's Digital Intelligence Technology Branch, announced the TeleChat model of China Telecom's ChatGPT product.
According to reports, the model supports outputting codes, writing speeches, etc. The company's research and development goal is to become a "tens of billions AI service provider", and its products cover AI algorithms, platforms, applications, hardware, large models, etc. Other information has not been announced yet.
Yunhan artificial intelligence team launched the first large language model "TourGPT" for the cultural tourism industry in China
Babbitt News, Yunhan artificial intelligence team released and launched China's first large language model "TourGPT" specially created for the cultural travel industry today, which will bring intelligent services and solutions to the domestic cultural travel industry.
Sichuan Changhong: "Changhong Super Brain" is compatible with AI large models such as HKUST Xunfei
According to Jiemian News, Sichuan Changhong recently stated on the interactive platform that at present, some Changhong TV products have been equipped with the "Changhong Super Brain" artificial intelligence system, which is officially open to users. "Changhong Super Brain" is compatible with AI large models such as HKUST Xunfei.
Ronglian Cloud releases "Chitu Large Model" and "Generative Intelligent Marketing Services Integrated Workspace"
According to 36 Krypton reports, on July 8, Ronglianyun officially released the vertical industry multi-level large language model "Chitu Large Model" for enterprise applications at the 2023 World Artificial Intelligence Conference, and released Ronglianyun's "generative intelligence The marketing service integrated workspace - Doraemon" and the "generative integrated intelligent customer service platform" aim to promote the digital transformation of enterprise marketing and services.
Peking University team released ChatLaw, a large legal model
According to Jinshi's report on July 4, the Peking University team recently released ChatLaw, a large Chinese legal model. Peking University Shenzhen Graduate School-Tuzhan Intelligence AIGC Joint Laboratory uses a large amount of structured text data in the legal field for training on the basis of a general large model, and recruits senior lawyers to assist in manual annotation and conduct high-quality fact-based multi-round dialogues to develop At the same time, it also open sourced 3 models: ChatLaw-13 B, ChatLaw-33 B and ChatLaw-TextVec. The project is committed to providing inclusive legal services, and currently has more than 1k Stars on GitHub.
Daguan Data officially released the "Cao Zhi" large model, and officially tested it to the outside world
According to Jinshi's report on July 7, during the 2023 World Artificial Intelligence Conference, the AI company Daguan Data officially released the "Cao Zhi" large model and opened it to the public for public testing. It is understood that Cao Zhi's large model supports functions such as automatic writing of long texts in multiple languages and multilingual translation, and fully empowers scenarios such as long document writing, machine translation, semantic analysis and review, knowledge question and answer, and text-to-SQL.
Enterprise Chacha Publisher Chacha Model "Knowing Enemy Alpha"
On July 3, Qichacha released a large model of Shangcha —— Qichacha "Knowing the Enemy Alpha". It is reported that this large model is based on the results of large-scale pre-training conducted by Enterprise Chacha after nearly ten years of credible data accumulation in the field of business queries. The innovative products that will be launched in the future will be jointly constructed through the large model + enterprise credit database, using AI technology to empower enterprise credit big data innovation, creating safe and credible artificial intelligence products, and providing users with more convenient and accurate business information query services.
Yuntian Lifei disclosed the large-scale model "Yuntian Book" for the first time
According to a report by "Kechuangban Daily" on July 6, at the 2023 World Artificial Intelligence Conference, Yuntian Lifei disclosed the large-scale model "Yuntian Book" for the first time. The large model consists of three levels: the general large model, the industry large model and the scene large model. In addition, Yuntian Lifei also demonstrated a new generation of edge computing chips Deep Edge10 series SoC chips. The chip will be successfully taped out at the end of 2022. It adopts domestic advanced technology and supports chiplet technology for multi-chip expansion.
application
The "Second Generation of Socks" in Zhuji, Zhejiang uses AI to personalize socks and sell them at double the price
According to a video released by "Beautiful Zhejiang" on Douyin, the "Second Generation of Socks" in Zhuji, Zhejiang uses AI to personalize socks and sell them at double the price. Zhou Yurun, general manager of Zhejiang Sanite Socks Co., Ltd., said in an interview, "Originally, foreign customers may place an order of 5 million pairs of identical socks at one time. Now a set of socks may change from hundreds of thousands of sets to There are thousands of sets, but the styles may have become hundreds. The original socks that cost two or three yuan can be sold for 6-8 yuan through customization. Socks are mainly exported to Canada and the United States, and some are Mexico."
NetEase Youdao’s self-developed educational model “Ziyue” presents a virtual human language coach for the first time
According to a report from Lieyun.com, on July 6, at the 2023 World Artificial Intelligence Conference, the latest application result of the first domestic large-scale education model "Ziyue" developed by Netease Youdao, the virtual human language coach, made its debut. During the conversation, this AI speaking coach demonstrated super strong reasoning ability, language ability and emotional ability, which attracted many audiences to stop.
Microsoft Global Senior Vice President: AI assistant Copilot has helped more than 1 million developers around the world complete nearly half of the code programming
According to the "Kechuang Board Daily" report on July 6, Hou Yang, the global senior vice president of Microsoft, chairman and CEO of Microsoft Greater China, said at today's 2023 Global Artificial Intelligence Conference that the company has launched GPT4 and enterprise-level The five major models including GPT support customers to create customized intelligent services for different industries, and at the same time bring mix365 commercial applications around Microsoft365 office, Top open source community development, digital information security protection, employee experience improvement, and Windows operating system. In addition, Microsoft has also launched a series of AI-driven Copilot smart assistant services. He said that since the launch of GitHub Copilot for programming assistance, more than 1 million developers around the world have used it, and nearly half of their code has been completed by Copilot, and the programming speed has increased by more than 50%. .
** Xueersi Online School launched the AIGC course "The First Lesson of Artificial Intelligence" based on the self-developed large model**
According to IT Home news, on the afternoon of July 3, at the Xueersi Online School Programming Product Upgrade and Artificial Intelligence Course Conference, Xueersi announced the launch of an AIGC (Generative Artificial Intelligence) course based on a self-developed large language model, called "Artificial Intelligence First Lesson". According to the official statement of Xueersi, the course will "promote the upgrading of programming course products and help popularize AI education".
Tsinghua University, together with Tencent and other institutions, launched the EEG image generation model DreamDiffusion
According to news from IT House on July 5, researchers from Shenzhen International Graduate School of Tsinghua University, Tencent Artificial Intelligence Laboratory (AI Lab) and Pengcheng Laboratory recently published a paper saying that the research team developed an image generation method called DreamDiffusion. model that can generate high-quality images directly from electroencephalogram (EEG) signals. Related papers were published on arXiv, an online academic paper platform owned by Cornell University.
The Hungarian company launched the world's first AI-developed beverage, the taste and packaging of which are all completed by AI
According to IT House news, HELL ENERGY, a Hungarian functional drink manufacturer, announced on July 3 that the company has successfully developed the world's first functional drink developed entirely by AI. From shape design, formula development to taste evaluation, safety measures and even marketing elements of this beverage, every link is completed by artificial intelligence. It is reported that this beverage will be produced in the most advanced equipment in Europe and will be sold in more than 60 countries and regions around the world starting this summer.
Aliyun launched a new AI painting product "Tongyi Wanxiang"
According to 36 Krypton's report on July 7, at the 2023 World Artificial Intelligence Conference, Alibaba Cloud officially launched a new AI painting product "Tongyi Wanxiang". Tongyi Wanxiang is built on the basis of Alibaba Cloud's "Tongyi" large model capability. Users can input prompt words in Tongyi Wanxiang to output corresponding images. In addition to Wen Shengtu, Tongyi Wanxiang has also launched functions including style transfer and similarity diagram generation. At present, Tongyi Wanxiang has been officially launched to provide external services.
Jingdong’s Yanxi model has written 3 billion words of marketing copy and tens of thousands of live broadcast scripts
According to a report by IT House on July 6, at the 2023 World Artificial Intelligence Conference today, He Xiaodong, vice president of Jingdong Group and director of the Exploration Research Institute, said that large models will reshape all business models, and Jingdong is the forefront of artificial intelligence development. Possess industry-leading AI technology capabilities. At present, JD.com's Yanxi model has written 3 billion words of marketing copy and tens of thousands of live broadcast scripts.
Supervision
The United States will restrict Chinese users from using Amazon, Microsoft, etc. to provide cloud services for training AI models
According to Reuters, citing the Wall Street Journal, the U.S. government is preparing to introduce a series of measures to restrict Chinese companies and individuals from using U.S. cloud computing services in order to protect U.S. data security and technological advantages. These measures may include export licensing requirements for cloud service providers, background checks for cloud service users, regulation of cloud service content, etc.
If the new regulation is passed, it may require US cloud service providers such as Amazon and Microsoft to obtain government permission before they can provide cloud computing services related to AI chips in China. The Commerce Department is expected to implement the restrictions in the coming weeks as part of an expansion of semiconductor export controls introduced in October.
policy
Shanghai issued a large-scale model policy to create an AI "model" city
According to China Business News, at the closing ceremony of the World Artificial Intelligence Conference (WAIC) held on the afternoon of July 8, Shanghai announced "Several Measures to Promote the Innovation and Development of Artificial Intelligence Large-scale Models in Shanghai" and released the "Model City" initiative. Three batches of artificial intelligence projects were signed, and the Shanghai artificial intelligence open source ecological industry cluster was established. According to Zhang Ying, deputy director of the Shanghai Economic and Information Commission, the measures focus on the four major directions of innovation capability, innovation elements, innovation applications, and innovation environment, focusing on building three plans and five major platforms. Three plans to break the bottleneck of development, the first is the large-scale model innovation support plan, which focuses on supporting Shanghai's innovative entities to develop large-scale models with international competitiveness, implements special rewards, and accelerates model iteration; the second is the intelligent computing power acceleration plan to strengthen large-scale models The third is the demonstration application promotion plan to strengthen the in-depth application and benchmarking scenarios of large models in vertical fields such as intelligent manufacturing, education and teaching, technology and finance, design creativity, and scientific intelligence.
Shanghai Xuhui released a new policy to support generative artificial intelligence, supporting independent innovation in key areas of large models
According to a report by the Financial Associated Press on July 7, at the 2023 World Artificial Intelligence Conference (WAIC) theme forum on "General Artificial Intelligence Industry Development Opportunities and Risks in the Era of Large-scale Models", "Several Measures of Xuhui District to Support the Development of Generative Artificial Intelligence "Official release. Yu Linwei, member of the Standing Committee of the Shanghai Xuhui District Committee and deputy head of the district, pointed out that Xu Huihui first launched support policies for generative artificial intelligence, mainly focusing on strong technical support, supporting independent innovation in key areas such as general-purpose large-scale models and vertical large-scale models, and supporting embodied intelligence. Focus on scene application demonstrations, give full play to Xuhui's characteristic industrial foundation and first-mover advantages in new tracks, and comprehensively promote the integration and empowerment of generative artificial intelligence technology in fields such as medical care, finance, social networking, and intelligent networked vehicles. capabilities; focus on the construction of an innovative ecosystem, and strengthen the support of elements for key links such as computing power, data, algorithms, financing, talents, and supervision.
View
Jingdong He Xiaodong: The two-month cost of large-scale model training is expected to be tens of millions of RMB, and he is very confident in the business prospects
According to The Paper reported on July 7, He Xiaodong, vice president of Jingdong Group and director of the Exploration Research Institute, said in an interview with reporters that the current basic general-purpose large model with a training time of about two months is estimated to cost tens of millions of yuan. The business prospects and landing scenarios of the large model are very confident. He suggested that start-up companies should find their own "moat" when they enter the big model. Facing the current situation of "100 model wars", He Xiaodong believes that for the market, pressure and competition are good things, which will effectively promote the development of the industry.
Gao Wen, Academician of the Chinese Academy of Engineering: Building a large model without a large computing power is a fantasy
According to Fenghuang.com, on July 7, Gao Wen, academician of the Chinese Academy of Engineering and director of Pengcheng Laboratory, said in a speech at the World Artificial Intelligence Conference Ascension Artificial Intelligence Industry Summit Forum that it is impossible to build a large model without large computing power. , This is the same as electricity. If you can save some electricity to make any products, other people who really know how to do it will not believe it. You must have computing power as the basis.
British Prime Minister Sunak: AI has great potential and is expected to bring 10% benefit to UK GDP in the next ten years
According to a report by Jinshi on July 4, British Prime Minister Sunak emphasized in his speech the great potential of artificial intelligence (AI), including the possibility of curing incurable diseases such as cancer and dementia, as well as the potential of AI in the fields of economy, medical care and education. widely used. AI is seen as a general-purpose technology that increases productivity, but requires commensurate investment to make it happen. In this regard, he cites a recent report by PwC, which predicts that AI will bring a 10% benefit to the UK's gross domestic product (GDP) in the next decade.
Stability AI CEO: In 5 years, there will be no human programmers
According to a Decrypt report on July 4, Stability AI founder and CEO Emad Mostaque talked about the current state of the artificial intelligence industry and his vision in an interview. He stated that “within five years, the human programmer will be completely gone.” Mostaque concludes that the future does not look bright for human programmers, and there is some evidence to support his thesis. According to GitHub, “41% of all code is currently generated by AI.” Even more interestingly, “In 3 months, AI has surpassed Bitcoin and Ethereum in popularity on GitHub.”
Tencent Tang Daosheng: Large-scale industry model is a better path for enterprises to embrace AI
According to Jinshi's report on July 6, at the 2023 World Artificial Intelligence Conference Industry Forum, Tang Daosheng, Senior Executive Vice President of Tencent and CEO of Tencent Cloud and Smart Industry Business Group, believed that the fundamental goal of artificial intelligence development is to land in the industry, and the industry will The model will open the "golden age" of industrial upgrading. Tang Daosheng revealed that Tencent Cloud MaaS announced not long ago will be upgraded again in the near future.
Data shows that ChatGPT traffic dropped for the first time in June
According to news from IT House on July 6, ChatGPT’s popularity has shown signs of weakening. According to data from the analysis company Similarweb, the reading traffic and visitor data of the ChatGPT website declined for the first time in June this year. According to the disclosed data, the global traffic of the ChatGPT website in June this year decreased by 9.7% compared with the previous month, the number of visitors decreased by 5.7%, and the average time spent by visitors on the Internet decreased by 8.5%, but ChatGPT is still as " "No. 1 in the industry" has a prominent advantage over other competing products.
Li Qiang, vice president of Tencent: Industry models will become the tipping point of the value of AI large models
According to a report by "Kechuangban Daily" on July 7, at the Tencent Forum of the 2023 World Artificial Intelligence Conference, Li Qiang, vice president of Tencent and president of government and enterprise business, said that in the era of large models, computing power, network, and data constitute the foundation of the underlying infrastructure. "Iron Triangle", in addition to providing MaaS services, Tencent is also providing infrastructure services such as HCC high-performance computing clusters, Xingmai high-performance computing networks, and vector databases for the creation of large-scale industry models through cloud services. Li Qiang also believes that general-purpose large models are not the only direction for model application, and models for vertical industries will become the tipping point of the value of large models. With the continuous iteration of technology, the industrial application of large models will also usher in an acceleration.
Research Report
Report: In 2024, China will have a multilingual general model comparable to GPT-4
According to 36 Krypton's report on July 7, at the 2023 World Artificial Intelligence Conference (WAIC) Qiming Venture Capital Forum "Generative AI and Large Models: Change and Innovation", Qiming Venture Partners and Unfinished Research jointly released the report "Generative AI 》| State of Generative AI 2023. According to the report, in 2024 China will have a multilingual general-purpose model comparable to GPT-4; Long Context will lead the next breakthrough in LLM technology.
Report: In the first half of the year, global venture capital investment has shrunk by nearly half, and more than US$40 billion has flowed to AI startups
According to a report by IT House on July 6, data recently released by research firm PitchBook shows that global venture capital funding will almost halve in the first six months of 2023, highlighting a lack of investor enthusiasm and a sharp rise in interest rates that has reduced financing needs.
Despite the success of OpenAI's ChatGPT in sparking huge interest in artificial intelligence (AI) startups, global venture capital funding fell 48% to $173.9 billion in the first half of the year, with the number of deals also down 19%, the report showed. But AI startups are still hot. Global venture capitalists have poured more than $40 billion into AI startups in the past six months, including Microsoft's $10 billion investment in OpenAI and $1.3 billion in Inflection AI.
CITIC Securities: The development process and market space of the big language model in the 2B field are significantly better than the 2C market
According to a research report by CITIC Securities on July 7, the popularity of ChatGPT has declined significantly recently. According to SimilarWeb data, the global user visits in June fell by -9.7% month-on-month, and the market therefore has some doubts about the current wave of AI driven by large language models. According to the analysis of CITIC Securities, the decline in ChatGPT user traffic is mainly due to the decrease in freshness of individual users, and OpenAI’s active adjustment of operating strategies (computing resource cost constraints), and shifting its focus to the 2B field. The most intuitive evidence is: the current ChatGPT related plug-ins , The development tool chain continues to maintain high popularity, and at the same time, the products of most European and American software companies have begun to fully integrate generative AI.
Zhou Zhifeng, Partner of Qiming Venture Capital: American AI talents are in enterprises, while Chinese AI talents are in universities
On July 7, at the 2023 World Artificial Intelligence Conference, Zhou Zhifeng, a partner of Qiming Venture Capital, shared that there is a big difference in the distribution of generative AI talents between China and the United States. In the United States, 19% are in Google, followed by Microsoft, META, etc. , mainly for enterprises. And 12% of China's talents are in Tsinghua University, followed by Zhejiang University and Peking University. Among the top ten rankings, there is only one company, Alibaba, which ranks fifth, and the rest are universities.
View Original
The content is for reference only, not a solicitation or offer. No investment, tax, or legal advice provided. See Disclaimer for more risks disclosure.
One week of AIGC 丨 Not afraid of the US blockade, in 2024 China will have a multilingual general model comparable to GPT-4
Financing
Beilian Zhuguan completed the angel round financing of 50 million yuan, and its business covers the AI computing power market
According to ChinaVenture.com, recently, Zhejiang Beilian Zhuguan, a cloud resource management service provider, announced the completion of an angel round of financing of 50 million yuan, invested by Yuanjing Capital, Sequoia China Seed Fund and Zhouxuan Equity.
Shanda Network CEO Chen Tianqiao invested another 1 billion yuan to support AI brain science
According to a report by the Financial Associated Press on July 6, Tianqiao Brain Science Research Institute (TCCI) announced that it will invest another 1 billion yuan to support AI brain development at a forum themed on "Brain-Computer Intelligence Digital Life" at the Shanghai World Artificial Intelligence Conference (WAIC). Science, including the establishment of MindX laboratory, the construction of a series of cutting-edge laboratories such as AI + sleep dreamland, AI + anti-aging, each with a start-up fund of 50 million yuan, and will continue to invest.
AI start-up company "Fourth Paradigm" has been approved by the China Securities Regulatory Commission for IPO in Hong Kong
According to an announcement on the official website of the China Securities Regulatory Commission, Beijing Fourth Paradigm Technology Co., Ltd., a domestic AI start-up company, has completed the relevant procedural work for an IPO on the Hong Kong Stock Exchange and has been approved by the China Securities Regulatory Commission. It is expected to conduct a listing hearing in the near future.
big model
Data: The number of domestic large-scale models has reached 93, and the proportion of Beijing has dropped to 45% for the first time
On July 4th, according to the statistics of the WeChat public account "Towards the Future", the number of domestic large-scale models has reached 93, which is about to exceed 100, and the proportion of Beijing has dropped to 45% for the first time.
More than 30 large models will be unveiled at the World Artificial Intelligence Conference
According to Times Finance, at the 6th World Artificial Intelligence Conference that just passed, more than 30 large-scale models were unveiled, and more than 20 large-scale model forums were overwhelmed, and the relevant booths were crowded with visitors without exception. Among the more than 30 large models exhibited at the venue, there are many GM models from major manufacturers such as Baidu Wenxin Yiyan, Alibaba Tongyi Qianwen, Huawei Pangu Large Model, HKUST Xunfei Xinghuo Cognition, and Shangtang Ririxin Sensenova. large model. There are also vertical large-scale models launched by small and medium-sized manufacturers such as Lanzhou Technology, Lianyuan Technology, Mobvoi, Xinghuan Technology, Midu, and Daguan Data for segmented scenarios.
Gao Wen, academician of the Chinese Academy of Engineering: will build a 200 billion parameter natural language large model base, and its performance will be benchmarked against ChatGPT
According to a report by IT House on July 6, Gao Wen, academician of the Chinese Academy of Engineering and director of Pengcheng Laboratory, said at the WAIC2023 Shengteng Artificial Intelligence Industry Summit Forum that Pengcheng Laboratory has launched a large-scale brain model project, aiming to build the first domestic Completely independent innovation, open source and open natural language pre-training large model base, the parameter level reaches 200 billion, and the performance is benchmarked against ChatGPT.
OpenAI fully opens the GPT-4 API and discards some old Completions API models
Babbitt News, on July 7, OpenAI announced on its official website that the GPT-4 API has been fully opened for use, and all paid API users can directly access GPT-4 in 8K context without any waiting. OpenAI is expected to open the GPT-4 API to new developers by the end of July. OpenAI also introduced GPT-3.5 Turbo, DALL·E and Whisper API. Fine-tuned versions of the safe and reliable GPT-4 and GPT-3.5 Turbo will also be available, expected later this year.
Shanghai Artificial Intelligence Laboratory released a general large-scale model system for scholars
According to the "Kechuangban Daily" report on July 6, the Shanghai Artificial Intelligence Laboratory launched a general large-scale model system for scholars at the 2023 World Artificial Intelligence Conference Science Frontiers Plenary Session, including three bases: multimodal, Puyu, and Tianji. At the same time, it launched the first full-chain open source system for the development and application of large models.
Among them, the multi-modal large model has 20 billion parameters, supports 3.5 million semantic tags, and is world-leading in 80+ tasks; the Pu language large model is the first large model officially released in China with 100 billion parameters supporting multiple languages; The model is the world's first city-level NeRF model.
Huawei Cloud Pangu Large Model 3.0 is officially released
According to a report by the Financial Associated Press on July 7, at the Huawei Developer Conference 2023 (Cloud) held this afternoon, Huawei Executive Director and Huawei Cloud CEO Zhang Pingan announced that Huawei Cloud Pangu Model 3.0 was officially released.
Zhang Pingan said that the Pangu Large Model 3.0 is divided into a three-layer structure of L0 basic model, L1 industry model, and L2 scene model, which will reshape thousands of industries. According to reports, this is a large model completely oriented to the industry. Among them, the Pangu L0 layer basic large model includes a natural language large model with hundreds of billions of parameters, Pangu multi-modal large model, visual large model, prediction large model, scientific computing large model, etc. The L1 layer is a large industry model, including government affairs , finance, manufacturing, mining, etc. The L2 layer is the scene model.
Tianlang releases three products including a general language model and a semantic search engine
According to the "Kechuangban Daily" report on July 8, at the World Artificial Intelligence Conference, Tianrang released the large-scale product ecology "Tiankang Xiaobai", which includes three products, namely a large language model, a semantic search engine and a development platform. It is understood that the Tianjie Xiaobai universal language large model provides ChatGPT-like capabilities through deep learning training with 186 billion parameters, and has the capabilities of multi-language dialogue interaction, knowledge question and answer, and logical reasoning. The semantic search engine enhances the generation ability of large models based on accurate facts, so that it can output based on accurate facts and avoid misleading information that large models may produce.
China Mobile released two artificial intelligence industry models
According to a report by the Financial Associated Press on July 8, China Mobile officially released the Jiutian Haisuan government affairs model and the Jiutian customer service model. It is reported that the Jiutian Haisuan government affairs model is an industry-oriented model for government affairs created by China Mobile based on the rich experience in digital government construction accumulated in recent years. Relying on the Jiutian Haisuan government affairs model, the government service system will have a powerful government affairs Item understanding ability, multi-dimensional information association ability, and multi-interaction ability for complex issues and complex processes; Jiutian·Customer Service Large Model can analyze user questions and provide answers based on natural language descriptions provided by users, and can also cooperate with human customer service , analyze the semantics and context of historical communication content, summarize and summarize the key points and key information of the dialogue, and prompt and reply suggestions for the artificial customer service agent according to the dialogue content, so as to reduce the response time and form a three-way communication field of "big model-manual agent-user" , greatly improving the work efficiency of traditional manual customer service.
China Telecom announced the TeleChat large model, claiming that it will establish a "tens of billions AI service provider"
According to a report from IT House on July 3, at the Artificial Intelligence Summit Forum of the 2023 Global Digital Economy Conference held yesterday, Zhang Xin, deputy general manager of China Telecom Group's Digital Intelligence Technology Branch, announced the TeleChat model of China Telecom's ChatGPT product.
According to reports, the model supports outputting codes, writing speeches, etc. The company's research and development goal is to become a "tens of billions AI service provider", and its products cover AI algorithms, platforms, applications, hardware, large models, etc. Other information has not been announced yet.
Yunhan artificial intelligence team launched the first large language model "TourGPT" for the cultural tourism industry in China
Babbitt News, Yunhan artificial intelligence team released and launched China's first large language model "TourGPT" specially created for the cultural travel industry today, which will bring intelligent services and solutions to the domestic cultural travel industry.
Sichuan Changhong: "Changhong Super Brain" is compatible with AI large models such as HKUST Xunfei
According to Jiemian News, Sichuan Changhong recently stated on the interactive platform that at present, some Changhong TV products have been equipped with the "Changhong Super Brain" artificial intelligence system, which is officially open to users. "Changhong Super Brain" is compatible with AI large models such as HKUST Xunfei.
Ronglian Cloud releases "Chitu Large Model" and "Generative Intelligent Marketing Services Integrated Workspace"
According to 36 Krypton reports, on July 8, Ronglianyun officially released the vertical industry multi-level large language model "Chitu Large Model" for enterprise applications at the 2023 World Artificial Intelligence Conference, and released Ronglianyun's "generative intelligence The marketing service integrated workspace - Doraemon" and the "generative integrated intelligent customer service platform" aim to promote the digital transformation of enterprise marketing and services.
Peking University team released ChatLaw, a large legal model
According to Jinshi's report on July 4, the Peking University team recently released ChatLaw, a large Chinese legal model. Peking University Shenzhen Graduate School-Tuzhan Intelligence AIGC Joint Laboratory uses a large amount of structured text data in the legal field for training on the basis of a general large model, and recruits senior lawyers to assist in manual annotation and conduct high-quality fact-based multi-round dialogues to develop At the same time, it also open sourced 3 models: ChatLaw-13 B, ChatLaw-33 B and ChatLaw-TextVec. The project is committed to providing inclusive legal services, and currently has more than 1k Stars on GitHub.
Daguan Data officially released the "Cao Zhi" large model, and officially tested it to the outside world
According to Jinshi's report on July 7, during the 2023 World Artificial Intelligence Conference, the AI company Daguan Data officially released the "Cao Zhi" large model and opened it to the public for public testing. It is understood that Cao Zhi's large model supports functions such as automatic writing of long texts in multiple languages and multilingual translation, and fully empowers scenarios such as long document writing, machine translation, semantic analysis and review, knowledge question and answer, and text-to-SQL.
Enterprise Chacha Publisher Chacha Model "Knowing Enemy Alpha"
On July 3, Qichacha released a large model of Shangcha —— Qichacha "Knowing the Enemy Alpha". It is reported that this large model is based on the results of large-scale pre-training conducted by Enterprise Chacha after nearly ten years of credible data accumulation in the field of business queries. The innovative products that will be launched in the future will be jointly constructed through the large model + enterprise credit database, using AI technology to empower enterprise credit big data innovation, creating safe and credible artificial intelligence products, and providing users with more convenient and accurate business information query services.
Yuntian Lifei disclosed the large-scale model "Yuntian Book" for the first time
According to a report by "Kechuangban Daily" on July 6, at the 2023 World Artificial Intelligence Conference, Yuntian Lifei disclosed the large-scale model "Yuntian Book" for the first time. The large model consists of three levels: the general large model, the industry large model and the scene large model. In addition, Yuntian Lifei also demonstrated a new generation of edge computing chips Deep Edge10 series SoC chips. The chip will be successfully taped out at the end of 2022. It adopts domestic advanced technology and supports chiplet technology for multi-chip expansion.
application
The "Second Generation of Socks" in Zhuji, Zhejiang uses AI to personalize socks and sell them at double the price
According to a video released by "Beautiful Zhejiang" on Douyin, the "Second Generation of Socks" in Zhuji, Zhejiang uses AI to personalize socks and sell them at double the price. Zhou Yurun, general manager of Zhejiang Sanite Socks Co., Ltd., said in an interview, "Originally, foreign customers may place an order of 5 million pairs of identical socks at one time. Now a set of socks may change from hundreds of thousands of sets to There are thousands of sets, but the styles may have become hundreds. The original socks that cost two or three yuan can be sold for 6-8 yuan through customization. Socks are mainly exported to Canada and the United States, and some are Mexico."
NetEase Youdao’s self-developed educational model “Ziyue” presents a virtual human language coach for the first time
According to a report from Lieyun.com, on July 6, at the 2023 World Artificial Intelligence Conference, the latest application result of the first domestic large-scale education model "Ziyue" developed by Netease Youdao, the virtual human language coach, made its debut. During the conversation, this AI speaking coach demonstrated super strong reasoning ability, language ability and emotional ability, which attracted many audiences to stop.
Microsoft Global Senior Vice President: AI assistant Copilot has helped more than 1 million developers around the world complete nearly half of the code programming
According to the "Kechuang Board Daily" report on July 6, Hou Yang, the global senior vice president of Microsoft, chairman and CEO of Microsoft Greater China, said at today's 2023 Global Artificial Intelligence Conference that the company has launched GPT4 and enterprise-level The five major models including GPT support customers to create customized intelligent services for different industries, and at the same time bring mix365 commercial applications around Microsoft365 office, Top open source community development, digital information security protection, employee experience improvement, and Windows operating system. In addition, Microsoft has also launched a series of AI-driven Copilot smart assistant services. He said that since the launch of GitHub Copilot for programming assistance, more than 1 million developers around the world have used it, and nearly half of their code has been completed by Copilot, and the programming speed has increased by more than 50%. .
** Xueersi Online School launched the AIGC course "The First Lesson of Artificial Intelligence" based on the self-developed large model**
According to IT Home news, on the afternoon of July 3, at the Xueersi Online School Programming Product Upgrade and Artificial Intelligence Course Conference, Xueersi announced the launch of an AIGC (Generative Artificial Intelligence) course based on a self-developed large language model, called "Artificial Intelligence First Lesson". According to the official statement of Xueersi, the course will "promote the upgrading of programming course products and help popularize AI education".
Tsinghua University, together with Tencent and other institutions, launched the EEG image generation model DreamDiffusion
According to news from IT House on July 5, researchers from Shenzhen International Graduate School of Tsinghua University, Tencent Artificial Intelligence Laboratory (AI Lab) and Pengcheng Laboratory recently published a paper saying that the research team developed an image generation method called DreamDiffusion. model that can generate high-quality images directly from electroencephalogram (EEG) signals. Related papers were published on arXiv, an online academic paper platform owned by Cornell University.
The Hungarian company launched the world's first AI-developed beverage, the taste and packaging of which are all completed by AI
According to IT House news, HELL ENERGY, a Hungarian functional drink manufacturer, announced on July 3 that the company has successfully developed the world's first functional drink developed entirely by AI. From shape design, formula development to taste evaluation, safety measures and even marketing elements of this beverage, every link is completed by artificial intelligence. It is reported that this beverage will be produced in the most advanced equipment in Europe and will be sold in more than 60 countries and regions around the world starting this summer.
Aliyun launched a new AI painting product "Tongyi Wanxiang"
According to 36 Krypton's report on July 7, at the 2023 World Artificial Intelligence Conference, Alibaba Cloud officially launched a new AI painting product "Tongyi Wanxiang". Tongyi Wanxiang is built on the basis of Alibaba Cloud's "Tongyi" large model capability. Users can input prompt words in Tongyi Wanxiang to output corresponding images. In addition to Wen Shengtu, Tongyi Wanxiang has also launched functions including style transfer and similarity diagram generation. At present, Tongyi Wanxiang has been officially launched to provide external services.
Jingdong’s Yanxi model has written 3 billion words of marketing copy and tens of thousands of live broadcast scripts
According to a report by IT House on July 6, at the 2023 World Artificial Intelligence Conference today, He Xiaodong, vice president of Jingdong Group and director of the Exploration Research Institute, said that large models will reshape all business models, and Jingdong is the forefront of artificial intelligence development. Possess industry-leading AI technology capabilities. At present, JD.com's Yanxi model has written 3 billion words of marketing copy and tens of thousands of live broadcast scripts.
Supervision
The United States will restrict Chinese users from using Amazon, Microsoft, etc. to provide cloud services for training AI models
According to Reuters, citing the Wall Street Journal, the U.S. government is preparing to introduce a series of measures to restrict Chinese companies and individuals from using U.S. cloud computing services in order to protect U.S. data security and technological advantages. These measures may include export licensing requirements for cloud service providers, background checks for cloud service users, regulation of cloud service content, etc.
If the new regulation is passed, it may require US cloud service providers such as Amazon and Microsoft to obtain government permission before they can provide cloud computing services related to AI chips in China. The Commerce Department is expected to implement the restrictions in the coming weeks as part of an expansion of semiconductor export controls introduced in October.
policy
Shanghai issued a large-scale model policy to create an AI "model" city
According to China Business News, at the closing ceremony of the World Artificial Intelligence Conference (WAIC) held on the afternoon of July 8, Shanghai announced "Several Measures to Promote the Innovation and Development of Artificial Intelligence Large-scale Models in Shanghai" and released the "Model City" initiative. Three batches of artificial intelligence projects were signed, and the Shanghai artificial intelligence open source ecological industry cluster was established. According to Zhang Ying, deputy director of the Shanghai Economic and Information Commission, the measures focus on the four major directions of innovation capability, innovation elements, innovation applications, and innovation environment, focusing on building three plans and five major platforms. Three plans to break the bottleneck of development, the first is the large-scale model innovation support plan, which focuses on supporting Shanghai's innovative entities to develop large-scale models with international competitiveness, implements special rewards, and accelerates model iteration; the second is the intelligent computing power acceleration plan to strengthen large-scale models The third is the demonstration application promotion plan to strengthen the in-depth application and benchmarking scenarios of large models in vertical fields such as intelligent manufacturing, education and teaching, technology and finance, design creativity, and scientific intelligence.
Shanghai Xuhui released a new policy to support generative artificial intelligence, supporting independent innovation in key areas of large models
According to a report by the Financial Associated Press on July 7, at the 2023 World Artificial Intelligence Conference (WAIC) theme forum on "General Artificial Intelligence Industry Development Opportunities and Risks in the Era of Large-scale Models", "Several Measures of Xuhui District to Support the Development of Generative Artificial Intelligence "Official release. Yu Linwei, member of the Standing Committee of the Shanghai Xuhui District Committee and deputy head of the district, pointed out that Xu Huihui first launched support policies for generative artificial intelligence, mainly focusing on strong technical support, supporting independent innovation in key areas such as general-purpose large-scale models and vertical large-scale models, and supporting embodied intelligence. Focus on scene application demonstrations, give full play to Xuhui's characteristic industrial foundation and first-mover advantages in new tracks, and comprehensively promote the integration and empowerment of generative artificial intelligence technology in fields such as medical care, finance, social networking, and intelligent networked vehicles. capabilities; focus on the construction of an innovative ecosystem, and strengthen the support of elements for key links such as computing power, data, algorithms, financing, talents, and supervision.
View
Jingdong He Xiaodong: The two-month cost of large-scale model training is expected to be tens of millions of RMB, and he is very confident in the business prospects
According to The Paper reported on July 7, He Xiaodong, vice president of Jingdong Group and director of the Exploration Research Institute, said in an interview with reporters that the current basic general-purpose large model with a training time of about two months is estimated to cost tens of millions of yuan. The business prospects and landing scenarios of the large model are very confident. He suggested that start-up companies should find their own "moat" when they enter the big model. Facing the current situation of "100 model wars", He Xiaodong believes that for the market, pressure and competition are good things, which will effectively promote the development of the industry.
Gao Wen, Academician of the Chinese Academy of Engineering: Building a large model without a large computing power is a fantasy
According to Fenghuang.com, on July 7, Gao Wen, academician of the Chinese Academy of Engineering and director of Pengcheng Laboratory, said in a speech at the World Artificial Intelligence Conference Ascension Artificial Intelligence Industry Summit Forum that it is impossible to build a large model without large computing power. , This is the same as electricity. If you can save some electricity to make any products, other people who really know how to do it will not believe it. You must have computing power as the basis.
British Prime Minister Sunak: AI has great potential and is expected to bring 10% benefit to UK GDP in the next ten years
According to a report by Jinshi on July 4, British Prime Minister Sunak emphasized in his speech the great potential of artificial intelligence (AI), including the possibility of curing incurable diseases such as cancer and dementia, as well as the potential of AI in the fields of economy, medical care and education. widely used. AI is seen as a general-purpose technology that increases productivity, but requires commensurate investment to make it happen. In this regard, he cites a recent report by PwC, which predicts that AI will bring a 10% benefit to the UK's gross domestic product (GDP) in the next decade.
Stability AI CEO: In 5 years, there will be no human programmers
According to a Decrypt report on July 4, Stability AI founder and CEO Emad Mostaque talked about the current state of the artificial intelligence industry and his vision in an interview. He stated that “within five years, the human programmer will be completely gone.” Mostaque concludes that the future does not look bright for human programmers, and there is some evidence to support his thesis. According to GitHub, “41% of all code is currently generated by AI.” Even more interestingly, “In 3 months, AI has surpassed Bitcoin and Ethereum in popularity on GitHub.”
Tencent Tang Daosheng: Large-scale industry model is a better path for enterprises to embrace AI
According to Jinshi's report on July 6, at the 2023 World Artificial Intelligence Conference Industry Forum, Tang Daosheng, Senior Executive Vice President of Tencent and CEO of Tencent Cloud and Smart Industry Business Group, believed that the fundamental goal of artificial intelligence development is to land in the industry, and the industry will The model will open the "golden age" of industrial upgrading. Tang Daosheng revealed that Tencent Cloud MaaS announced not long ago will be upgraded again in the near future.
Data shows that ChatGPT traffic dropped for the first time in June
According to news from IT House on July 6, ChatGPT’s popularity has shown signs of weakening. According to data from the analysis company Similarweb, the reading traffic and visitor data of the ChatGPT website declined for the first time in June this year. According to the disclosed data, the global traffic of the ChatGPT website in June this year decreased by 9.7% compared with the previous month, the number of visitors decreased by 5.7%, and the average time spent by visitors on the Internet decreased by 8.5%, but ChatGPT is still as " "No. 1 in the industry" has a prominent advantage over other competing products.
Li Qiang, vice president of Tencent: Industry models will become the tipping point of the value of AI large models
According to a report by "Kechuangban Daily" on July 7, at the Tencent Forum of the 2023 World Artificial Intelligence Conference, Li Qiang, vice president of Tencent and president of government and enterprise business, said that in the era of large models, computing power, network, and data constitute the foundation of the underlying infrastructure. "Iron Triangle", in addition to providing MaaS services, Tencent is also providing infrastructure services such as HCC high-performance computing clusters, Xingmai high-performance computing networks, and vector databases for the creation of large-scale industry models through cloud services. Li Qiang also believes that general-purpose large models are not the only direction for model application, and models for vertical industries will become the tipping point of the value of large models. With the continuous iteration of technology, the industrial application of large models will also usher in an acceleration.
Research Report
Report: In 2024, China will have a multilingual general model comparable to GPT-4
According to 36 Krypton's report on July 7, at the 2023 World Artificial Intelligence Conference (WAIC) Qiming Venture Capital Forum "Generative AI and Large Models: Change and Innovation", Qiming Venture Partners and Unfinished Research jointly released the report "Generative AI 》| State of Generative AI 2023. According to the report, in 2024 China will have a multilingual general-purpose model comparable to GPT-4; Long Context will lead the next breakthrough in LLM technology.
Report: In the first half of the year, global venture capital investment has shrunk by nearly half, and more than US$40 billion has flowed to AI startups
According to a report by IT House on July 6, data recently released by research firm PitchBook shows that global venture capital funding will almost halve in the first six months of 2023, highlighting a lack of investor enthusiasm and a sharp rise in interest rates that has reduced financing needs.
Despite the success of OpenAI's ChatGPT in sparking huge interest in artificial intelligence (AI) startups, global venture capital funding fell 48% to $173.9 billion in the first half of the year, with the number of deals also down 19%, the report showed. But AI startups are still hot. Global venture capitalists have poured more than $40 billion into AI startups in the past six months, including Microsoft's $10 billion investment in OpenAI and $1.3 billion in Inflection AI.
CITIC Securities: The development process and market space of the big language model in the 2B field are significantly better than the 2C market
According to a research report by CITIC Securities on July 7, the popularity of ChatGPT has declined significantly recently. According to SimilarWeb data, the global user visits in June fell by -9.7% month-on-month, and the market therefore has some doubts about the current wave of AI driven by large language models. According to the analysis of CITIC Securities, the decline in ChatGPT user traffic is mainly due to the decrease in freshness of individual users, and OpenAI’s active adjustment of operating strategies (computing resource cost constraints), and shifting its focus to the 2B field. The most intuitive evidence is: the current ChatGPT related plug-ins , The development tool chain continues to maintain high popularity, and at the same time, the products of most European and American software companies have begun to fully integrate generative AI.
Zhou Zhifeng, Partner of Qiming Venture Capital: American AI talents are in enterprises, while Chinese AI talents are in universities
On July 7, at the 2023 World Artificial Intelligence Conference, Zhou Zhifeng, a partner of Qiming Venture Capital, shared that there is a big difference in the distribution of generative AI talents between China and the United States. In the United States, 19% are in Google, followed by Microsoft, META, etc. , mainly for enterprises. And 12% of China's talents are in Tsinghua University, followed by Zhejiang University and Peking University. Among the top ten rankings, there is only one company, Alibaba, which ranks fifth, and the rest are universities.