One week of AIGC丨It’s too cumbersome, Stability AI release is unnecessary_Drawing tool, Midjourney nuclear bomb level update

Last week, the two overseas AI painting giants Midjourney and Stable Diffusion were updated almost at the same time. Stable Diffusion XL 0.9 is released. The pictures generated by the new version are more realistic and textured, and can correctly generate human five fingers. Midjourney 5.2 is shockingly released. It can zoom like a camera, fill in the details of the picture, and customize the style. The boundless universe animation made with zoom out is directly shocking to the point of speechlessness.

Image source: Generated by Unbounded AI tool

Financing

Haitian Ruisheng: It is planned to raise no more than 790 million yuan, which will be used for AI large model training data set construction projects and data production vertical large model research and development projects

According to a report from Wall Street on June 21, Haitian Ruisheng stated that it plans to raise no more than 790 million yuan in additional funds, which will be used for AI large-scale model training data set construction projects and data production vertical large-scale model research and development projects.

Insai Group: It plans to raise no more than 645 million yuan, which will be used for marketing AIGC large-scale models and other projects

According to 36 Krypton’s report on June 21, Insai Group issued an announcement stating that the company plans to issue shares to specific objects to raise no more than 645 million yuan, which will be used for marketing AIGC large-scale model research and development and application projects, and supplementing working capital.

Great Aim Group established the AIGC Application Research Institute and the AIGC Special Fund, with an initial scale of 500 million yuan

According to the official Weibo news of Huace Film and Television, on the evening of June 18, Huace Group announced the establishment of the AIGC Application Research Institute and the establishment of a special fund for AIGC. Huace Film and Television AIGC Application Research Institute has President Fu Binxing concurrently as the dean, and Shen Xiong as the full-time vice dean. It has a system development department, a creative empowerment department, an application promotion department, and a development cooperation department. The AIGC special fund established this time has an initial scale of 500 million yuan, all of which are owned by Huace Film and Television. It will focus on vertical large-scale model companies based on the film and television industry, AI application companies in the film and television industry, film and television AI technology platforms and teams, etc. .

Y Combinator has invested in 88 AI-related projects this year: enterprise services and software development accounted for 75%

According to the "Kechuang Board Daily" report on June 20, according to the statistics of China National Finance Securities, Y Combinator, a well-known American business incubator, will participate in 270 projects in 2023, and 88 AI-related projects were selected from them, covering enterprise services, software development, education, Financial, entertainment, medical and other industries. Among them, the two major tracks of enterprise services and software development are the most popular, accounting for about 3/4 of the number of projects. Combining technical feasibility and cost-effectiveness of investment, the three main AI project landing paths are: provide model customization deployment services; replace part of manpower in the field of static knowledge to help enterprises reduce costs and increase efficiency; the field of enterprise services is a typical static knowledge application scenario, and it is also One of the most popular tracks for AI applications to land.

“Timely Language” received tens of millions of RMB Pre-A + round of financing, the funds are mainly used for cross-border AI customer service product development and iterative optimization

According to 36 Krypton’s report on June 26, “Timely Language” has completed tens of millions of yuan in Pre-A + round of financing, which is exclusively invested by Yuandu Capital. This round of financing will be mainly used for cross-border AI customer service product development and iterative optimization, and the establishment of regional sales teams. According to reports, Shijiyu was established in 2021, and has previously completed the Angel round and Pre-A round of investment from 01 Venture Capital and Jushuitan. Relying on AI technology, the company focuses on providing a multi-platform/multi-channel centralized and unified intelligent customer service SaaS platform for cross-border e-commerce sellers, aiming to meet the needs of overseas enterprise channel acquisition, customer operation and brand management.

AI + cloud industrial software company "Sanweijia" completed the D round of financing of hundreds of millions of yuan, with a post-investment valuation of nearly 600 million US dollars

According to 36 Krypton’s report on June 19, the AI + cloud industrial software company “Sanweijia” completed the D round of hundreds of millions of yuan in financing. The investors are Guoke Investment and Inovance Technology. The post-investment valuation will reach nearly 600 million US dollars. This round of financing will be mainly used to accelerate the iterative upgrade of its product matrix, and increase investment in research and development of technologies such as AI artificial intelligence, which will help 3D Home to focus more on the three major natural technologies of cloud modeling, cloud rendering, and 3D vector AI. Research and develop the core engine to strengthen Sanweijia's leading position in the field of home furnishing design, marketing and production software.

Global business information SaaS AI platform "AroundDeal" completed tens of millions of Pre-A round of financing

According to 36 Krypton’s report on June 20, “AroundDeal” announced that it has completed tens of millions of yuan in the Pre-A round of financing, and the investor in this round is becoming capital. The financing will be used for product development and marketing. According to reports, "AroundDeal", established in 2020, is an AI-powered global business information SaaS platform dedicated to providing business information to Chinese B2B companies going overseas and B2B companies in overseas emerging markets through AI and data technology.

AI speech generation platform ElevenLabs completes $19 million in Series A financing, with a valuation of $99 million

According to the "Kechuang Board Daily" report, on June 20, ElevenLabs, an AI speech generation platform, announced the completion of a US$19 million Series A financing. This round of financing was jointly led by entrepreneurs Nat Friedman and Daniel Gross, and A16Z (Andreessen Horowitz). Participants included Creator Ventures, SV Angel, Instagram co-founder Mike Krieger, Oculus co-founder Brendan Iribe, Deepmind and InflectionAI co-founder Mustafa Suleyman, and more. After the funding round, ElevenLabs is valued at $99 million and the startup is just over a year old, sources said.

AI Web3 Search Engine Kaito Completes $5.5M Series A Funding Led by Superscrypt and Spartan

According to CoinDesk, on June 22, Kaito, an artificial intelligence-based Web3 search engine, announced the completion of a US$5.5 million Series A financing, led by Superscrypt and Spartan, with a valuation of US$87.5 million. It is reported that Kaito is creating a Web3 search engine based on a large language model, and opened a beta version to 35,000 waiting list users on June 20.

AI pharmaceutical company Aureka completes $10 million seed round financing

According to the "Kechuangban Daily" report on June 25, AI pharmaceutical company Aureka Biotechnologies completed a seed round of financing of tens of millions of dollars jointly invested by Xianfeng Cowin and Newley Capital. After this round of financing, Aureka will further accelerate the construction of its AI+ high-throughput digital biotechnology macromolecular drug development platform and the research pace of finding general principles for drug design.

AI Reasoning Deployment Solution Provider "Qingyue Intelligence" Completed Tens of Millions of Angel Round Financing

According to 36 Krypton reports, on June 25, AI reasoning deployment solution provider “Qingyue Intelligence” announced the completion of an angel round of financing of tens of millions of yuan. This round of financing is exclusively led by Oasis Capital, and the financing funds will be mainly used for algorithm research and development, product development and team expansion. According to reports, at present, Qingyuan Intelligence has launched MLGuider, a reasoning optimization tool chain for AI models, to adapt and optimize the current mainstream hardware environment.

Entrepreneurship

Wang Huiwen, the co-founder of Meituan, left his post and was admitted to the hospital for medical treatment. He has been investing in large-scale entrepreneurship for more than 4 months

According to the "Kechuangban Daily" report, it was reported today that Wang Huiwen, the co-founder of Meituan and the co-founder of Light Years Beyond, recently encountered personal health problems and is currently away from work and admitted to the hospital for medical treatment. "Kechuangban Daily" learned from multiple sources that Wang Huiwen is currently seeking medical treatment in the hospital, and the specific cause is unknown. This afternoon, Lai Binqiang, the former HRD of Meituan, also said in Moments that he is accompanying Wang Huiwen in the hospital, and said that please rest assured that in the next period of time, he will need to rest and recover according to the doctor's advice, and everything will be fine. It has been established for more than 4 months, and the current valuation has reached the unicorn in the field of AI large-scale models.

Informed sources: Jacob Devlin, a well-known AI researcher, returned to Google after briefly quitting OpenAI

Well-known artificial intelligence researcher Jacob Devlin has returned to his old job, according to a person familiar with the matter, The Information reported. In January, Devlin left Google to join rival OpenAI after complaining internally about how Google trained its chatbot software, Bard AI.

big model

Tencent discloses the progress of large-scale models in the industry: do not pursue high parameters, hope that the cost budget can be controlled

According to The Paper, on June 19, Wu Yunsheng, vice president of Tencent Cloud and head of Tencent Cloud Intelligence, said in an interview, “At present, the combination of large models and industries is still in the early stage. The understanding of the large model is also deepening.” Wu Yun claimed that he hopes to make the large model really well with the lowest cost, the most suitable model and the best service, and will provide the best model under the condition of budget and cost control. A logical model, let it achieve its purpose in the scene and solve the problem. Wu Yunsheng also revealed that Tencent Cloud's large-scale industry model capabilities have been implemented in various products such as Tencent Qidian, Tencent Conference, and Tencent Cloud AI Code Assistant.

Ant Group is developing a large model, named "Zhenyi"

According to the "Kechuangban Daily" report on June 21, the technology research and development team of Ant Group is developing its own language and multi-modal large-scale model, which is named "Zhenyi" internally.

Jingdong announced that the AI model will be released on July 13

According to news from IT House on June 21, JD.com officially announced today that the "2023 JD.com Global Technology Explorer Conference and JD.com Cloud Summit" will be held in Beijing on July 13th. According to reports, the theme of this conference is "Leaping Industrial Intelligence", focusing on the deep integration of large-scale models and industries. It will release JD large-scale models, launch a new generation of digital infrastructure, and upgrade products and solutions.

Model "Daily New": SenseTime's "SenseChat 2.0" comprehensive performance of multiple evaluation benchmarks exceeds ChatGPT

According to a report from Flush on June 21, SenseTime recently announced the results of its self-developed Chinese language large model "SenseChat 2.0" on the three authoritative large language model evaluation benchmarks of MMLU, AGI, and C-. According to the evaluation results, "SenseChat 2.0" performed better than ChatGPT in the three test sets, realizing an important breakthrough in the research of large language models in my country

Li Yanhong: The Wenxin Model 3.5 version has increased the training speed by 2 times, and the inference speed has increased by 17 times

According to Fenghuang.com news on June 26, Robin Li, founder, chairman and CEO of Baidu, attended the "Nishan Dialogue on Digital Civilization at the World Internet Conference" and delivered a speech entitled "Large Models Reshape the Digital World". Robin Li revealed in his speech that the Baidu Wenxin model has been iterated to version 3.5. Compared with the 3.0 version, the training speed has been increased by 2 times, the reasoning speed has been increased by 17 times, and the cumulative effect of the model has increased by more than 50%.

Li Yanhong believes that "the key point of the new international competition strategy is not how many large models a country has, but how many native AI applications are on your large models, and the extent to which these applications have improved production efficiency. If we can squeeze Go to the poker table and get tickets to the competition, China will have a stronger digital industry, and the scale of the digital economy will grow tremendously."

Changhong TV officially released the AI model "Changhong Super Brain"

According to news from IT House on June 20, at today's Changhong Artificial Intelligence Large Model TV Conference, the Changhong TV AI large model "Changhong Super Brain" was officially released. According to reports, Changhong Ultrain has three core capabilities of perception, understanding and creation. Realized the smart TV "the first step into the era of strong artificial intelligence". It can have a natural and continuous dialogue with users, and can also create content and services according to the individual needs of users.

AI company MosaicML launched a 30 billion parameter model MPT-30B, and said that the training cost is only a fraction of competing products

According to IT House's report on June 25, AI startup MosaicML recently released its language model MPT-30B. The model has 30 billion parameters, and the training cost is "only a fraction of other similar competing models." The training cost of such models expands the application of AI models in a wider range of fields.

application

Midjourney released version 5.2, zooming like a camera and filling the details of the picture

According to the heart of the machine, following the launch of version 5.1 last month, Midjourney quickly updated version 5.2. This update includes improved aesthetics, coherence generation, text understanding, and provides prompt word analysis, high-variation mode, and high-resolution mode. One of the most exciting is the new "Zoom out" function, through 1.5 times, 2 times and custom zoom, the camera is pulled out and all side details are filled to achieve the effect of reconstructing the image.

Stable Diffusion upgrade: Five fingers are generated normally, and the image is more realistic

According to news from IT House on June 24, Stability AI issued a press release a few days ago, announcing the launch of SDXL 0.9 version update, which upgraded the Stable Diffusion text generation image model.

The pictures generated by the new version are more realistic and textured, and can correctly generate human five fingers. Stability AI said that the SDXL 0.9 version configuration requirements are the same as before, requiring 16 GB of memory and NVIDIA RTX 20 series and higher graphics cards (requires 8 GB and above video memory), and supports Win10, Win11 and Linux distributions. Stability AI said that users can access the model through ClipDrop, and the related API will be launched soon, and version 1.0 is expected to be released in mid-July.

Meitu releases 7 AIGC products, including AI visual model, digital human, etc.

Babbitt News, at the Meitu Image Festival held today, Meitu (01357.HK) released 7 AIGC tools, including the AI visual model "MiracleVision", the digital human product "DreamAvatar", and the AI-inspired design tool " WHEE" etc. According to the "Science and Technology Innovation Board Daily", these 7 products are mainly biased towards the B-end market, and will be launched in internal testing in July.

**Kuaishou’s video editing product "Kuaiying" has internally tested a variety of AIGC creation functions, which can generate copywriting with one click, etc. **

According to Tech Planet’s report on June 20, recently, Kuaishou’s video editing product “Kuai Ying” and functional shooting product “Yitian Camera” have been tested in succession in the near future, including “AI animation video”, “AI copywriting recommendation ", "AI Instant Universe", "AI Stickers" and many other AIGC creation functions.

Opera, a subsidiary of Kunlun Wanwei, officially launched the AI browser Opera One

According to a PRNewswire report on June 20, Kunlun Wanwei's Web3 browser Opera announced that the AI browser Opera One has ended its beta testing and is officially launched. Users can now download it and currently support Windows, macOS and Linux operating systems.

Supervision

National Cyberspace Administration of China released a list of in-depth synthetic service algorithm filings, Baidu, Ali, Tencent Byte and other major manufacturers are listed

On June 20, the Cyberspace Administration of China released a domestic in-depth synthesis service algorithm filing list (June 2023), including Meituan Online Smart Customer Service Algorithm, Kuaishou Short Video Generation Synthesis Algorithm, Baidu Wenshengtu Content Generation Algorithm, Baidu PLATO Big Model algorithm, volcano engine intelligent beautification special effect algorithm, Tencent meeting virtual background algorithm, Tmall Xiaomi intelligent customer service algorithm, Xunfei speech recognition algorithm, etc.

Singapore government agency: currently does not plan to regulate artificial intelligence

Singapore is in no rush to formulate artificial intelligence regulations despite repeated calls for government intervention to address its risks, CNBC reported on June 19. Lee WanSie, director of trusted artificial intelligence and data at the Singapore Infocomm Media Development Agency, said that regulation of artificial intelligence is not currently considered, but efforts are being made to promote the responsible use of artificial intelligence, calling on companies to cooperate in the development of the world's first AI testing toolkit— AI Verify, which enables users to perform technical tests on their AI models and document process checks. AI Verify starts in 2022 as a pilot project. Tech giants IBM and Singapore Airlines have already started pilot tests as part of the plan.

The U.S. Department of Commerce will establish an AI public working group to deal with the potential risks of generative AI

According to IT House, U.S. Secretary of Commerce Gina Raimondo (Gina Raimondo) recently announced that the National Institute of Standards and Technology (NIST) will establish a new public working group on artificial intelligence (AI). It will build on the success of the NIST Artificial Intelligence Risk Management Framework (RMF) to further address issues brought about by the rapid development of AI technology.

Japan drafts guidelines for the use of generative AI for schools, prohibiting students from writing essays with ChatGPT

According to IT House citing Japan's "Yomiuri Shimbun", Japan's Ministry of Education, Culture, Sports, Science and Technology recently stated that it has drafted guidelines for schools for generative AI represented by ChatGPT, and will release them as soon as possible after listening to expert opinions.

The draft pointed out that generative AI brings various risks such as copyright infringement and dissemination of false information, so it is appropriate for schools to take the lead in trying to "restrict the use of generative AI." On this basis, the draft proposes specific "inappropriate" cases of the use of generative AI in schools, including the use of generative AI in literary creation activities such as poetry and haiku, or in examinations.

policy

Shanghai released an action plan to create a new track for cultural tourism metaverse, and the industrial scale will exceed 50 billion yuan by 2025

On June 21, the activity of creating a cultural and tourism metaverse track in Shanghai was held at the Opportunity Center of Shanghai Tomorrow Square. At the event, the Shanghai Municipal Bureau of Culture and Tourism released the "Shanghai Action Plan for Building a New Track for Cultural Tourism Metaverse (2023-2025)". The action plan proposes that by 2025, Shanghai's "Cultural Tourism Yuan Universe" brand will be initially launched, the level of development will be at the forefront of the country, the industrial scale will exceed 50 billion yuan, and 2-3 new content creation industry cluster development zones will be formed. Fang Shizhong, director of the Shanghai Municipal Bureau of Culture and Tourism, said that it is necessary to develop digital cultural tourism production tools, encourage the use of digital technology to improve content production efficiency, optimize production tools for modeling, driving, rendering and other links, and promote market entities to carry out AIGC (Artificial Intelligence Independently Generated Content ) tool platform development.

big company

Accenture signs $100M generative AI project in 4 months

According to a LiveMint report on June 25, Accenture said that in the past four months, the company signed $100 million worth of projects in the field of generative artificial intelligence. "Gen AI is a big opportunity, but it's still early days," said Accenture Chairman and CEO Julie Sweet.

Amazon invested 100 million US dollars to establish an AI innovation center to help AWS business development

According to news from IT House on June 24, Amazon announced this week that it will invest 100 million US dollars to establish an AI innovation center to assist enterprise customers and partners in planning and deploying AI technology. Amazon hopes to use this technology to promote its AI cloud services and catch up with competing products from industry rivals such as Microsoft and Google, so as to expand the attractiveness of its AWS business to enterprises and help AWS business development.

Tesla created a Twitter account "Tesla AI": the basic model of autonomous robots is being manufactured

On June 21, Tesla created a Twitter account for Tesla AI, stating that "Tesla is building the basic model for autonomous robots." In addition, Tesla CEO Musk also followed the account and forwarded several of the tweets.

View

Zuckerberg, Bill Gates and other four rich people made 1 trillion yuan because of AI, and AI became a "money printing machine" in seconds

According to statistics from Bloomberg, most of the top 10 richest people on the Bloomberg Rich List are from the technology sector. This wave of US stock market rebound led by AI will increase their total wealth by more than 400 billion US dollars (approximately RMB 2.86 trillion yuan). Among them, only Zuckerberg, Larry Ellison, Bill Gates, and Huang Renxun's total wealth increased by US$150 billion (equivalent to more than RMB 1 trillion).

Securities Daily: Listed companies AI "surf", beware of "chaos"

On June 22, an article in the Securities Daily pointed out that at present, the AI large-scale model technology represented by ChatGPT is still in its early stage, and there are many problems in itself; application still has a long way to go. For managers of listed companies, in the face of revolutionary technological breakthroughs in the field of AI, it is understandable to worry about being "anxious" caused by being left behind in the great changes, but they should also beware of "chaos" and avoid blindly following the trend and participating in the trend game on .

OpenAI CEO Responds to AI Concerns: Global Regulation Can Address Significant Risks, But Shouldn't Excessive

According to Jinshi's report on June 23, OpenAI CEO Sam Altman said when talking about the rapid rise of artificial intelligence technology, "There are many ways that can go wrong, and we are using dangerous technology that may often be used for dangerous purposes." Expressing concern over the possible damage from increasingly powerful artificial intelligence technologies, he said global regulation could address significant risks but should not overdo it. “I have enough money,” he said, when asked whether he would benefit financially from OpenAI’s success. His motivation is to exploit the potential benefits of the technology, hoping to contribute to human technological progress through artificial intelligence: "I think it's going to be the most important step that humanity has to make possible with technology, and I really care about that."

Research Report

Report: 43% of companies will deploy AI interviews next year

According to a report by IT House on June 20, according to a report released by the market research agency Resume Builder earlier this month, 43% of corporate recruitment in 2024 will introduce artificial intelligence to simplify the recruitment process and improve interview efficiency.

It is estimated in the report that 15% of enterprise personnel managers will adopt fully automatic recruitment methods, relying entirely on artificial intelligence to help enterprises screen suitable talents and conduct interviews with them. A recent survey also revealed that 46% of job seekers are using ChatGPT to create their application materials. Interestingly, some 78% of respondents said companies had higher response rates and interview invites after using AI technology in the job search process.

TrendForce: It is estimated that the total capacity of AI accelerator chips equipped with HBM in the market will reach 290 million GB in 2023

According to the "Kechuangban Daily" report on June 21, TrendForce Research pointed out that in 2023, chatbot and other generative AI applications will drive the growth of AI servers, and large cloud operators will be the most active in investing, including Microsoft, Google, AWS, Baidu, ByteDance, etc. have successively purchased high-end AI servers to continuously train and optimize their AI analysis models. The high-end AI chips needed for high-end AI servers will boost the demand for high-bandwidth memory (HBM) in 2023-2024. It is estimated that in 2023, the market will mainly be equipped with HBM's AI acceleration chips (such as NVIDIA's H100 and A100, AMD's MI200 and MI300, and Google's TPU, etc.), and the total capacity of HBM will reach 290 million GB, with a growth rate of nearly 60%. It will continue to grow by more than 30% each year.

Report: U.S. generative AI jobs surged 20% in May

According to a report by IT House on June 23, according to the latest data released by the recruitment and job search website Indeed, the number of jobs related to generative AI in the United States increased by 20% in May this year from the previous month, indicating that more and more American companies hope to catch the wind. Layout or use this technology to promote enterprise development. According to data from May, there are 204 generative AI-related jobs for every million job postings in the United States, which is more than double that of 2021. Among Indeed's artificial intelligence positions, data scientist positions account for 5%, and positions such as software engineers, machine learning engineers, and data engineers are also popular among companies.