One week AIGC 丨Wang Xiaochuan created the Chinese version of OpenAI, and Ali's version of ChatGPT was launched for testing

The intelligent era will once again open the "Renaissance". Everyone will be stronger because of AI. With good AI, a person may be able to become a company, solve problems that were difficult to solve before, and become the person they want to be. Society will also be flatter, and intellectual services previously available to a few people will be more inclusive. ——Wang Xiaochuan, CEO of Baichuan Intelligence

Image source: Generated by Unbounded AI Tools

financing

Wang Xiaochuan's Chinese version of OpenAI "Baichuan Intelligence" has received an investment of $50 million

On April 10, according to Xin Zhiyuan, Wang Xiaochuan officially announced his entry into China's big model entrepreneurship, the new company named "Baichuan Intelligence", has convened an elite team of 50 people, received 50 million US dollars of start-up capital, and is expected to release "China's best big model" by the end of the year.

Portrait Analytics raised $3 million for generative AI investment research platform

On April 6, Finextra reported that Portrait Analytics closed a $3 million seed round led by 406 Ventures. The company's first product is a question-and-answer-based application that combines generative AI search and summary for public market analysts. The technology responds to users' tasks by extracting and synthesizing key information buried in company documents, resulting in factual responses that can be fully audited by users.

Google-backed AI startup Anthropic plans to raise billions of dollars to train the model Claude-Next

On April 7, as TechCrunch reported, AI research startup Anthropic aims to raise up to $5 billion over the next two years to counter rival OpenAI and enter more than a dozen major industries. Anthropic's Series C funding filing also revealed other long-term goals for the company, which was founded in 2020 by Dario Amodei, former vice president of research at OpenAI. In its introduction, Anthropic said it plans to build a "cutting-edge model" — tentatively called "Claude-Next" — that is 10 times stronger than today's most powerful AI capabilities, but that would cost $1 billion over the next 18 months.

Cosmic Singularity completed a series A financing of 30 million yuan, which will be used in the field of AI

According to reports from the investment community, recently, Zhejiang Cosmic Singularity completed a series A financing of 30 million yuan, led by a leading domestic artificial intelligence AI application innovative technology company, and the old shareholders Chuangda Capital, Zhongwei Yihe Equity Fund followed. This round of funding will focus on the research and development of the next-generation Broadcast Genie AI Digital Human Quantum Server, the layout of the national distribution agent network, the acceleration of the popularization and application of the Broadcast Genie AI Digital Human combined with local life in short video and live broadcast scenarios, and the upgrading of AI technology stores in the whole industry.

Shenzhen "Mingxin Mathematics" completed a series A + financing of nearly 100 million yuan

According to a report by 36Kr on April 7, Tianyan inspection shows that Shenzhen Mingxin Digital Intelligence Technology Co., Ltd. (hereinafter referred to as "Mingxin Digital") recently completed a series A + financing of nearly 100 million yuan, and the investor is CDH Investment. Founded in 2017, "Mingxin Mathematics" is a financial technology company that builds AI models for industrial operation evaluation based on the foundation of industrial Internet big data.

Web3 social fitness app GoJoe completes £1.2 million seed funding

On April 3, Tech.eu reported that GoJoe, a social fitness app for Web3, completed a £1.2 million (about $1.5 million) seed round in funding, led by Superbet Ventures, with participation from Animoca Brands and former Manchester United footballer Michael Carrick. The round aims to expand its B2 B SaaS offering and further internationalize.

GoJoe is a social platform that aims to bring workplaces and individuals together through gamification and AI-based platforms to achieve their fitness goals.

technology

Alibaba's version of ChatGPT is available for testing

On April 7, the Ali version of ChatGPT, developed by Ali Damo Academy, Tongyi Qianwen, began inviting users to test the experience. At this stage, the model mainly invites enterprise users to conduct experience testing, and users can apply through the official website (tongyi.aliyun.com), and qualified users can participate in the experience. It is reported that Ali Damo Academy has been laying out cutting-edge scientific research fields such as NLP natural language processing for many years, and launched the research and development of large models in 2019.

Microsoft Research introduces a new AI ecosystem TaskMatrix.AI that connects foundational models with millions of APIs

According to MarktechPost on April 7, Microsoft researchers recently launched a new artificial intelligence ecosystem TaskMatrix.AI, allowing the integration of base models with millions of existing models and system APIs to form a "super artificial intelligence" capable of performing a variety of digital and physical tasks. TaskMatrix.AI is built on four main components that work together to enable the system to understand the user's goals and execute API-based executable code for specific tasks, including the Multimodal Dialogue Base Model (MCFM), API Platform, API Selector, and API Executor.

Glean introduces new generative AI capabilities to enhance enterprise search and discovery

April 7, VentureBeat reported that Glean, Inc., which provides intelligent search and discovery solutions for knowledge workers, announced a new set of capabilities on Tuesday that uses artificial intelligence (AI) to synthesize and surface relevant information across the organization. These capabilities are based on Glen's proprietary knowledge model and are designed to improve the accuracy and security of enterprise data and content, as well as employee productivity and collaboration, especially in remote or hybrid work environments.

apply

The high-resolution global weather forecast AI large model "Fengwu" was released, and the effective forecast time exceeded 10 days for the first time

According to Jin Shi reported on April 7, the Shanghai Artificial Intelligence Laboratory, together with the University of Science and Technology of China, Shanghai Jiao Tong University, Nanjing University of Information Science and Technology, the Institute of Atmospheric Physics of the Chinese Academy of Sciences and the Shanghai Central Meteorological Observatory, released the global medium-term weather forecast model "Fengwu". Based on multi-modal and multi-task deep learning methods, the AI large model "Fengwu" can achieve effective prediction of core atmospheric variables for more than 10 days at high resolution for the first time.

According to the Shanghai Artificial Intelligence Laboratory, "Fengwu" surpasses the meteorological large model GraphCast released by Google's DeepMind in 80% of the evaluation indicators. In addition, "Fengwu" can generate global high-precision forecasts for the next 10 days in only 30 seconds, which is much better than traditional models in terms of efficiency.

Unbounded AI launched the "First AI Comic Video" Contest to compete for the strongest "AI player" in 1 month

Babbitt, April 9, AIGC creation tool Unbounded AI launched a comic video contest with the theme of "Smart Art, Everyone Creates", the competition requires that original content (created by humans or AI) or licensed fiction script works as a model, using Unbounded AI to create a collection of comic works, and edited into a video submission, video comic content includes but is not limited to comic pages, four frames, commentary, graphics, dynamic comics, etc., but the final form needs to be uploaded by video. The contest runs from April 14 to May 14.

Affected by the outbreak of the AI industry, the market size of general-purpose GPUs is expected to exceed $34.557 billion in 2027

Babbitt on April 9, domestic cloud computing related experts believe that the minimum threshold for computing power to do a good AI large model is 10,000 NVIDIA A100 chips. TrendForce research shows that based on the computing power of A100, GPT-3.5 large models require up to 20,000 GPUs, and may need more than 30,000 in the future commercialization. In the long run, the development and deployment of large models in the future is an inevitable trend, and behind each large model training and deployment, there are tens of thousands of GPU chips supporting. Therefore, the market demand for general-purpose GPUs will usher in explosive growth. According to VerifiedMarket Research, the Chinese mainland discrete GPU market was valued at $4.739 billion in 2020 and is expected to exceed $34.557 billion in 2027.

Inspur Information: The demand for AI servers has exploded, and a larger liquid cooling base is being planned

On April 9, according to CaiLian News Agency, its reporter learned in a visit to the Inspur Information Tianchi liquid cooling industrial base that the continuous growth of AI computing power is one of the factors driving the demand for liquid cooling, Inspur information server product line people said that "AIGC hot brought about an explosion in demand for AI servers, in March we found that the AI (server) market has grown several times, previously (customer concern) is whether it can be cheaper, now can it be used." It is reported that the annual production capacity of the Tianchi base exceeds 100,000 units, and the company is also planning a larger liquid cooling production base; In terms of cost, the company expects that the "wind liquid parity" is expected to be realized in 2025.

BGI: AI applications can be introduced and developed in many aspects of the medical field

On April 9, according to Golden Ten Data, BGI said in an institutional survey that at present, BGI's synthetic production platform has diversified solutions, organically combining iBEST difficulty gene synthesis technology with other technologies, which can achieve rapid delivery of batch short genes, overcome long difficulty genes, and expand the scope of business delivery. AIGC technology is a combination of artificial intelligence and graph computing, which can assist us in intelligent analysis and processing on massive amounts of data to indicate disease risk and select appropriate treatments. AI applications can be introduced and developed from tool development, data production, information analysis, and knowledge discovery.

Yingyu Group has been connected to GPT3.5 turbo, and its products will be upgraded and explored around AIGC in the future

On April 4, according to titanium media reports, Yingyu Group (formerly known as Yingke Interactive Entertainment Group) has been connected to GPT3.5 turbo, and GPT4 is under review. At the same time, Yingyu Group also recently announced its layout plan in AIGC. The relevant person in charge said that the exploration of the future universe in the field of AIGC mainly has five major directions: digital human live broadcast, interactive social networking, AI music, AI script and AI operation. For example, in the field of interactive social networking, in the future, it will better interact with users by accessing GPT4 and become a "TA" who understands users better. At the level of AI operation layer, AI provides rich design creativity for products and produces illustrations that fit the scene. AI scripts are expressed as rewriting novels, generating novel outlines, and working with screenwriters to complete works.

Large companies

ChatGPT temporarily closed Plus payment due to too much demand

On April 5, according to Wall Street News, ChatGPT has now closed Plus payment, and OpenAI gave the reason that the demand is too large. Previously, ChatGPT caused heated discussions due to large-scale bans, and the signed letter led by Musk and others to suspend the development of large models above GPT4 has also attracted great attention in the industry.

OpenAI CEO talks with Japanese prime minister: consider opening a Japan office and expanding Japanese language services

After talks with Japanese Prime Minister Fumio Kishida, Sam Altman, co-founder and CEO of OpenAI, said the company is considering opening a Japanese office and expanding Japanese language services, as reported by Kim Shi on April 10. "We want to spend more time interacting with great people and building great products for the Japanese people,"

Baidu issued a statement against counterfeiting: At present, Wenxin has no official app, and has launched a lawsuit against Apple and related developer teams

News on April 8, "Baidu AI" WeChat public account issued a statement saying: At present, Wenxin Yiyan does not have any official App, before the company's official announcement, all "Wenxin Yiyan" apps seen in the App Store and various app stores are fake, and all those who claim to have "Wenxin Yiyan source code" are scammers. The Company has filed a lawsuit against Apple and its developer team in the Haidian District People's Court in Beijing.

Meta says generative AI may be used to create virtual worlds in the metaverse

Meta, the parent company of Facebook and Instagram, intends to launch its exclusive generative AI tool by the end of this year, and in addition to helping enterprise customers create ads, Meta also hopes that generative AI can eventually be used to create virtual worlds in the metaverse, as reported by Decrypt on April 6. It is reported that the Web3 metaverse platform Oncyber is already testing this feature.

PepsiCo: Has invested hundreds of millions of dollars in AI technology to integrate AI with sales, product research and development, etc

On April 7, Athina Kanioura, executive vice president and chief strategy and transformation officer of PepsiCo, said that since she joined the company, it has begun to integrate AI with sales, product research and development and other processes, and PepsiCo has spent hundreds of millions of dollars to invest in AI technology. AI helps companies better understand consumer preferences and has successfully shortened product development cycles from years to months.

Government position

Canada begins investigation of OpenAI over data security issues

On April 5, according to Kim Shi, citing Agence France-Presse, the Office of the Privacy Commissioner of Canada (OPC) announced on the 4th that it began investigating OpenAI, the company behind ChatGPT, involving complaints that "allege OpenAI collecting, using and disclosing personal information without consent". The Italian Data Protection Authority (Garanteperla Protezionedei DatiPersonali) temporarily banned ChatGPT late last month and launched an investigation into the tool's alleged breach of privacy rules.

Japan may consider adopting AI technologies such as ChatGPT in government work

As reported by Kim 10 on April 10, Japan's Chief Cabinet Secretary Hirokazu Matsuno said that if privacy and cybersecurity issues are addressed, the adoption of AI technologies, including ChatGPT, will be considered in government work.

The EU will require all generative AI content to be sourced

On April 3, according to CaiLian News Agency, in response to the chaos caused by AI-generated pictures, Thierry Breton, the EU's internal market commissioner and head of industry and information industry, publicly stated that the EU wants mandatory regulations on AI-generated content. "The EU will be the first government to make a clear stipulation that everything generated by AI, whether text or images, will be obliged to inform that it is done by AI," Breton said. "The EU Artificial Intelligence Bill will be officially voted on from this month, and generative AI will also be included as a regulatory priority." If someone posts an AI-generated image on a social network that doesn't indicate that it was AI-generated, they will face legal action. Breton said.

point of view

Wu Jun: ChatGPT is not a new technological revolution and does not bring new opportunities

On the evening of April 3, Wu Jun, a computer scientist and natural language model expert, was invited to the live broadcast room to launch a live broadcast on hot topics such as artificial intelligence and ChatGPT. Wu Jun said that historically ChatGPT is not actually a technological revolution, it affects the lazy person, too lazy to use his brain, to create new things. Those who truly explore the mysteries of human knowledge will not be replaced.

"King of Mathematics" Chengtong Yau: AI will not change the social structure in ten years

According to a report by Jin Shi on April 7, Chengtong Yau, the first Fields Medal Chinese winner known as the "king of mathematics", said that "artificial intelligence cannot have any impact on the most cutting-edge mathematicians, and there may be some impact on ordinary mathematicians." He also predicts that artificial intelligence will not have a big impact on the social structure of humans within a decade, and most of the talk of AI replacing humans is "alarmist."

Huawei Yuntianqi: It takes 12 million US dollars to develop and train large models, and the market will be concentrated in large companies

As reported on April 8, Tian Qi, Chief Scientist of HUAWEI CLOUD Artificial Intelligence Field, delivered a speech entitled "AI Empowers Intelligent Upgrade in Thousands of Industries" at today's AI Big Model Technology Summit Forum. He said that the development of artificial intelligence has moved from local exploration to thousands of industries, and AI has entered the core production system of enterprises and begun to create greater value. The penetration rate of the AI industry has accelerated, reaching 20% by 2026. He shared two major trends facing the development of artificial intelligence: First, from small models to large models, the computing power demand of AI algorithms has increased by 400,000 times in the past 10 years. Large models may incorporate highly customized small models, resulting in market concentration to large companies: high capital threshold, large model development and training of 12 million US dollars at a time; The technical threshold is high, and the AI framework is deeply optimized and the parallel computing capability is super strong.

Pan Yunhe, academician of the Chinese Academy of Engineering: AIGC will turn to more valuable application areas and suggest timely layout

According to a report by Jin Shi on April 8, the main forum of the "Artificial Intelligence Large Model Technology Summit Forum" hosted by the Chinese Intelligence Society opened in Xiaoshan, Hangzhou. Pan Yunhe, academician of the Chinese Academy of Engineering, said that AIGC is an important big concept, in practice, only based on the support of big data, big knowledge and big computing power, AIGC has a wide range of development space and application fields, and the large model can provide strong support for AIGC, which in turn has a huge impact on economic and social development. AIGC will definitely not be used only for chatting and drawing, but will move on to more valuable applications, and it is recommended to layout in time. Such as: AIGC of the real economy, intelligent design generation of new products, new processes, and new drugs; AIGC for culture and art, such as advertising, animation, film and television, painting, music, intelligent content generation for children's education; AIGC for urban and rural development, such as urban planning, beautiful countryside, online meetings, ecological deduction, etc. Intelligent simulation generation.

Research report

Report: Generative AI like ChatGPT Leads to 135% Increase in Phishing Email Attacks

On April 4, CSOonline reported that attackers using generative AI such as ChatGPT increased the volume of social engineering attacks by 135% by increasing text descriptions, punctuation, and sentence length, according to a new research report published by cybersecurity firm Darktrace. The report looked at more than 6,700 employees in the UK, US, France, Germany, Australia and the Netherlands, and 82% were concerned that hackers could use generative AI to create scam emails that were indistinguishable from real communications.

CITIC Securities: Continue to be optimistic about investment opportunities in the global AI field

According to a report by CaiLian News on April 7, CITIC Securities pointed out that the good demonstration effect brought by ChatGPT in the global AI industry and technology field is expected to promote the AI algorithm model to end the current technical route fork, and continue to move closer to GPT-led large language model (LLM), accelerating the arrival of the "industrialization" era of the global AI industry. As one of the most certain industrial directions in the medium term, we continue to be optimistic about investment opportunities in the global AI field, and recommend continuing to focus on core links such as chips, computing power facilities, model architecture & engineering practices, and application scenarios.

Industrial Securities: The Next Step of AI Investment, Focusing on "Data Elements + State-owned Cloud"

On April 5, according to interface news reports, Industrial Securities Research Report believes that AI investment should pay more attention to data in addition to computing power and algorithms. Various large model manufacturers are gradually launching vertical industry applications, and policies are also accelerating the digitalization of the industry. Under the background of vertical scenario data, data security, and autonomous and controllable AI large models, AI investment should pay more attention to "data elements + state-owned cloud". In summary, "data element operators" with operation and governance government affairs data, as well as data in key industries and key vertical fields, as well as independent and controllable state-owned cloud vendors with the background of central state-owned enterprises, are expected to usher in value revaluation.

Huatai Securities: The financial industry is expected to become one of the first vertical fields for AI large models to land

According to a report by 36Kr on April 6, Huatai Securities believes that the financial industry is leading in digitalization and has a data bonanza, which is expected to become one of the vertical fields where AI large models are the first to land. On the application side, we see that generative and comprehensible big models are landing or have potential landing scenarios in many sub-fields such as banking, insurance, asset management, investment research, and investment advisory, helping financial institutions reduce costs and increase efficiency.

Zheshang Securities: A new round of scientific and technological revolution with artificial intelligence as the core is the most important driving force for medium- and long-term economic growth

reported on April 6, Zheshang Securities Research Report believes that a new round of scientific and technological revolution with artificial intelligence as the core is the most important driving force for medium- and long-term economic growth and an important direction for the rebuilding of the mainland's industrial base. The core of "AI+" is industrial intelligence, and in the process of industrial intelligence, it is necessary to pay more attention to the new opportunities brought by the change of production and life paradigm: First, the change of production paradigm, digital elements may become important assets in traditional industries. Second, supply creates demand, the process of industrial intelligence will give birth to new scenarios, new demand, bring about people's life paradigm change, and reshape various traditional lifestyles and related industry tracks such as clothing, food, housing, medical care and pension, it is recommended to pay attention to seven major tracks: smart cars, smart homes, smart home appliances, smart buildings, smart medical care, smart pension and smart education.

Soochow Securities: AI space is vast, optimistic about financial data related targets

According to a report by Jin Shi on April 10, Soochow Securities pointed out in the research report that AI is comparable to the fourth technological revolution, and the rapid breaking of ChatGPT has promoted the continuous interpretation of the AI market; In the era of AI, industry leaders with more financial data will take the lead in landing financial GPT, and the successful landing of BloombergGPT shows the deep integration of large models in the financial field. Under the background of insufficient model scale, the feasibility of the scheme to achieve capability enhancement through high-quality vertical data has been proved. On the whole, considering the valuation cost performance and thematic correlation, in the direction of AI, it is recommended to pay attention to leading companies that master rich financial vertical data and layout AI.