AIGC has evolved again. Google announced at the 2023 I/O Developers Conference that the MusicLM model will be open to all users, and music can be generated by entering text descriptions. The epic upgrade of GPT4's strongest opponent, Anthropic Claude, is summarized in one minute with a hundred pages of information. Stability AI has launched the Stable Animation SDK, a text-to-animation tool that allows developers and artists to generate animations using the Stable Diffusion model. Wen Sheng pictures, Wen Sheng music, Wen Sheng animation, read a 10,000-character long text in 1 minute, what else can't be done by AIGC?
Image source: Generated by Unbounded AI tool
01
financing
Blockchain and metaverse media company Yaoshi Xinghui signed a $60 million strategic investment agreement with two institutions
According to PRNewswire, Nasdaq-listed blockchain and metaverse media company Glory Start announced that it has signed a $60 million strategic investment agreement with two institutions at a price of $2.48 per share. The transaction is expected to be completed in 2023 Finished around May 9th.
Haitu Information received more than 10 million yuan in angel round financing, and the funds will be mainly used for AIGC core technology and equipment research and development
According to a report by the Financial Associated Press on May 8, recently, Haitu Information recently completed an angel round of financing of over 10 million yuan, jointly invested by Wuhai Capital and Qianhai Kunpeng Capital. This round of funds will be mainly used for AIGC core technology and equipment research and development, high-level talent building, marketing and expansion. Haiguo Tuzhi is an intelligent service platform based on AIGC technology, providing various information for 60 million cross-border small and medium-sized enterprises, including AI live broadcast marketing, daily selected business intelligence, in-depth consultation and other intelligent services in the industrial chain.
AI pharmaceutical company Yulu Ganxing completed the second round of financing of tens of millions of yuan
According to the "Kechuangban Daily" report on May 8, the AI pharmaceutical company "Yulu Qianxing" has recently completed the second round of financing of tens of millions of yuan, led by Fujian Suzhou Fund under Fosun Pharma, followed by Zhongxin Capital. cast. The financing will be mainly used for the iteration of the company's core algorithm and computing power improvement and large-scale market expansion.
AI image generation technology service provider "Tiamat" received a million-dollar A+ round of investment
According to 36 Krypton’s report on May 8, AI image generation technology service provider “Tiamat” recently received millions of dollars in A + round investment, led by Linear Capital, followed by old shareholders DCM and Oasis Capital. Vision Capital acted as the exclusive financial advisor for this round of financing. At present, Tiamat has accumulated active small and medium-sized enterprise customers in industries such as FMCG, advertising, fashion, and design. In the future, Tiamat will continue to iterate its own research and development capabilities and polish its product system.
AI detection tool GPTZero receives $3.5 million in seed funding and will launch a new browser plugin
AI detection tool GPTZero has raised $3.5 million in seed funding from early-stage venture firms Uncork Capital and Neo, as well as Altman Capital and Stability AI CEO Emad Mostaque, Forbes reported. Reuters CEO Tom Glocer and former New York Times CEO Mark Thompson also participated in the round. In addition, GPTZero plans to soon launch a browser add-on called Origin to assess the accuracy and provenance of facts in a published article.
AI+Web3 medical technology company Lavita AI completed a $5 million seed round of financing
According to Crypto Reporter, AI+Web3 medical technology company Lavita AI announced the completion of a $5 million seed round of financing, led by Camford Capital, and a group of angel investors from the blockchain and artificial intelligence fields participated. The new financing will be used to accelerate the development of Lavita's AI-first platform and applications for patients, healthcare providers and research institutions.
AI company Going out to ask for an IPO in Hong Kong within this year at the earliest, or raising funds of 200 million to 300 million US dollars
According to Jinshi citing informed sources, Chinese artificial intelligence company and smart device manufacturer Momenwen is cooperating with CICC and CMB International to prepare for a Hong Kong IPO that may be launched within the year as soon as possible, and the scale of financing may reach 200 million to $300 million.
02
large model
OpenAI is using GPT-4 to explain the behavior of large language models
According to Siliconangle, OpenAI is using GPT-4 to explain the behavior of large language models. Currently, the tool is far from complete, but the company has made the code open source and available on GitHub for others to explore and refine. In its research, OpenAI attempted to use GPT-4 to understand one of its predecessors, GPT-2.
Unisound will release the large model of Yunzhisheng Mountain and Sea on May 24
According to news from IT House on May 13, the domestic AI company Yunzhisheng announced that it will hold the Yunzhisheng Mountain and Sea Large Model and Achievement Conference on May 24, and will also release smart medical care, smart IoT and Products and solutions for business services.
Google released the AI language model "PaLM 2", and the fierce battle between AI manufacturers promotes the accelerated development of the industry
According to a report by the Financial Associated Press on May 11, Google CEO Sundar Pichai released its latest artificial intelligence (AI) language model "PaLM 2" at the Google I/O conference today, competing with OpenAI's GPT-4 and others. The opponent initiates a challenge. Google also said that PaLM 2 has been used to support 25 of its own features and products, including the AI chatbot Bard, as well as Google Docs, Slides and Sheets.
03
application
Google announced the opening of the MusicLM model to all users
According to news from IT House on May 11, Google launched the MusicLM model in January this year, which can automatically generate music works according to the text requirements entered by users. In a blog post released that day, Google announced that the MusicLM model is open to all users. It is reported that the MusicLM model can generate 5-minute songs based on text after 280,000 hours of audio data training.
The Anthropic Claude model is upgraded to 100,000 tokens, which can swallow the entire "The Great Gatsby" in one gulp
According to TechCrunch, artificial intelligence startup Anthropic announced that it has expanded the context window of its flagship text generation AI model Claude from 9,000 tokens to 100,000 tokens. Anthropic said that in addition to reading long texts, the upgraded Claude can help retrieve information from multiple files or even a book. Anthropic lists several possible use cases, including digesting, summarizing and interpreting documents such as financial statements or research papers; analyzing a company's risks and opportunities based on its annual report; evaluating the strengths and weaknesses of a piece of legislation; identifying risks, Argumentation on the subject and different forms of legal documents, etc.
OpenAI will roll out web browsing and plugins to all ChatGPT Plus users next week
Babbitt News, on May 13, OpenAI tweeted that we will roll out web browsing and plug-ins to all ChatGPT Plus users next week! From alpha to beta, ChatGPT will be allowed to access the Internet and use more than 70 third-party plugins.
Google releases Tailwind, an "AI notebook" learning tool project
According to Jiemian News, on May 10, local time, Google released the AI file management tool project "Project Tailwind" at its I/O Developer Conference, which was described as a future notebook and a private artificial intelligence model that can Trained on documents uploaded by users. Users can ask the AI questions and get feedback on documents. Google is pitching it as a learning tool, and it's still in the "early" stages, with waitlist testing only available in the US.
Stability AI launches Stable Animation SDK, a text-to-animation tool
According to Zdnet, on May 12, artificial intelligence company Stability AI announced the launch of a text-to-animation tool that allows developers and artists to generate animations using the Stable Diffusion model. The tool, called the Stable Animation SDK, can generate video from three different inputs, including plain text, text and splash image, and text and input video.
Vision China launched an AI creative tool that can re-create copyrighted images
Babbitt News, on May 8, Vision China announced that its website vcg.com launched the first AIGC application product - AI intelligent search, and launched a new AIGC function - a creative tool based on generative AI technology. Users can re-create copyrighted pictures online by using AI creative tools - model conversion and illustration conversion functions. Currently, AI creative tools are only available to some contracted customers. Visual China will soon launch more AI creative tools: screen expansion, intelligent image retouching; and AI content generation function inspiration drawing, commodity background synthesis, etc. According to reports, after the customer downloads the second-created picture and confirms the payment, the creator of the original copyright picture can still get the corresponding revenue.
Leo Shares: Subsidiary Leo Digital launched the AIGC platform "LEO AIAD"
According to Jiemian News on May 8, LEO Digital, a subsidiary of LEO, announced the launch of the AIGC platform "LEO AIAD" for the entire marketing industry. According to reports, at present, "LEO AIAD" can achieve a maximum output value of up to 720,000 pictures per day, and the speed of designers' drawing can be increased from the original average of 6 hours per picture to 1 minute per picture. The platform is open to all employees of the company.
Zhou Hongyi announced the launch of "360 AI Store", which will integrate global AI tools
According to the Securities Times e company report, on May 13, Zhou Hongyi, the founder of 360 Group, announced that 360 will launch an AI store. 360 AI Store, as an AI navigation, will integrate global AI tools. On the one hand, it will take advantage of traffic to provide a display window for startups developing AI vertical applications, and on the other hand, it will make it easier for ordinary users to use AI tools. Zhou Hongyi said that the big language model is developing by leaps and bounds, and China has entered the "ten thousand model war", and users should be allowed to "vote with their feet". At present, the 360 AI store brings together hundreds of tools in more than ten categories, such as AI drawing, AI writing, AI audio production, etc., for unified display and navigation.
04
Supervision
The Japanese government held the first "AI strategy meeting" to discuss the rules of use
According to a report by the Financial Associated Press on May 11, the Japanese government held the first meeting of the "AI Strategy Conference" at the prime minister's residence to discuss formulating rules for the use of artificial intelligence (AI). As generative AI such as the AI chat software "ChatGPT" is becoming more and more popular in society, the conference will discuss the ways in which educational institutions, administrations, and companies use AI and the regulatory mechanism. Prime Minister Fumio Kishida attended the meeting and pointed out: "AI has the potential and risks to make the economy and society more positive. It is necessary to properly deal with both. I hope that the content and results of the meeting will be reflected in the formulation of international rules."
Council of EU lawmakers agrees to draft tougher AI rules
Leading EU lawmakers agreed on Thursday to amend draft rules to rein in generative artificial intelligence such as ChatGPT and proposed a ban on facial surveillance, Reuters reported, bringing the bloc closer to setting landmark rules for artificial intelligence. The plenary will vote on the draft next month after the two parliamentary committees responsible for the issue reached an agreement, after which lawmakers will hammer out details with EU countries and the European Commission.
05
dispute
AI singer cover or involved in multiple infringements
According to a CCTV news report quoted by the Financial Associated Press, recently, through model training and post-processing, the "AI singer" who simulates the voice of a real singer for cover singing has become popular on the Internet. Experts suggest that "AI singer" may involve multiple infringement issues: the Civil Code clearly refers to the provisions of portrait rights to protect the voice of natural persons, and forgery is not allowed to infringe on other people's portrait rights; the copyrights of original songwriters, singers, and recording producers are also Can not be ignored. At present, the industry is discussing the development and management of generative artificial intelligence, and our country will soon have norms to follow.
06
research report
IDC: It is estimated that the scale of China's artificial intelligence market will exceed US$14.7 billion in 2023
According to Jinshi's report on May 12, IDC predicts that the scale of China's artificial intelligence market will exceed US$14.7 billion in 2023, and will exceed US$26.3 billion in 2026. Zhong Zhenshan, vice president of IDC China, believes that market growth will mainly come from the replacement of AI applications built in the past few years by applications based on large models, the incremental market brought by generative AI, and new AI-enabled enterprise-level applications. In the future, the general AI market in the past will become increasingly saturated, and it will be difficult for manufacturers without large-scale model capabilities to maintain their competitive advantage.
Soochow Securities: The film and television industry is a high-quality landing scenario for generative AI
According to a May 12 report from the "Kechuang Board Daily", Soochow Securities pointed out that the film and television industry, as a comprehensive audio, video, and text, is a high-quality landing scene for generative AI; and from the perspective of the value and path of AI applications, animated films It is expected to be the first to benefit.
Compared with the 1-2 year production cycle of live-action movies, animated movies take longer (3-4 years), the project is more complicated, and the production capacity constraints are more obvious. Generative AI technology is expected to empower animation film script creation, mid-term painting, final video editing/dubbing/video output and other links, break the bottleneck of animation film production capacity, and further open up market space by increasing the supply of high-quality animation films in the industry.
Minsheng Securities: The emergence of "miniaturization" and "offline" models, edge-side terminals and chip iterations are expected to accelerate
According to the "Science and Technology Innovation Board Daily" report on May 12, Minsheng Securities' latest research report pointed out that Google's latest PaLM2 includes four models: Gecko, Otter, Bison and Unicorn. Among them, Gecko, as the lightest version of PaLM2, can run quickly on the mobile device side, providing the ability to process 20 tokens per second. Gecko can provide AIGC with more customized, lower latency, and better security and privacy applications, and open up scenarios where AI models are reasoned on terminal chips. Edge-side smart terminals and SoCs are expected to be fully upgraded and enter a new update cycle. .
Babbitt Park is open for cooperation!
Chinese Twitter: https://twitter.com/8BTC_OFFICIAL
English Twitter: https://twitter.com/btcinchina
Discord community: https://discord.gg/defidao
Telegram channel: https://t.me/Mute_8btc
Telegram community: https://t.me/news_8btc
