Babbitt | Metaverse Daily Must-Read: Tencent announced the progress of large-scale model research, saying that it has provided more than 50 large-scale model industry solutions for more than 10 industries such as cultural tourism, government affairs, and finance

Abstract: According to reports from Wall Street, on June 19, Tencent Cloud held the industry large model and intelligent application technology summit. At the summit, Tencent Cloud announced the research progress of its large-scale industry model, and said that it will rely on the Tencent Cloud TI platform to build a selected store of large-scale industry models to provide customers with one-stop MaaS (Model-as-a-Service) services to help Customers build exclusive large models and intelligent applications. Tang Daosheng, Senior Executive Vice President of Tencent Group and CEO of the Cloud and Smart Industry Business Group, said that Tencent Cloud has provided more than 50 large-scale industry solutions for more than 10 industries such as cultural tourism, government affairs, and finance.

Image source: Generated by Unbounded AI tool

Hot info:

Tencent has provided large-scale model solutions for more than 50 industries: covering more than 10 industries such as cultural tourism, government affairs, and finance

According to Wall Street reports, on June 19, Tencent Cloud held an industry large-scale model and intelligent application technology summit at the National Science and Technology Communication Center. At the summit, Tencent Cloud announced the research progress of its large-scale industry model, and said that it will rely on the Tencent Cloud TI platform to build a selected store of large-scale industry models to provide customers with one-stop MaaS (Model-as-a-Service) services to help Customers build exclusive large models and intelligent applications.

At the summit, Tang Daosheng, Senior Executive Vice President of Tencent Group and CEO of the Cloud and Smart Industry Business Group, said that the industrial scene is the best training ground for large-scale models. A large model industry solution. Tang Daosheng also pointed out that the large model is only the beginning, and the integration of AI and industry will bloom a more creative future. Ecological co-construction is an effective path for AI development. Tencent will adhere to ecological openness, provide high-quality model services for enterprises, and support customers' multi-model training tasks to accelerate the innovation and exploration of large models in industrial scenarios.

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. According to the market, Meitu (01357.HK) rose in late trading. As of press time, the company's shares rose nearly 27%.

Singapore government agency: No plans to regulate artificial intelligence yet

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.

Princeton University launched the Infinigen AI model, which can generate real natural environment 3D scenes

According to a report from IT House on June 19, Alexander Raistrick, a researcher at Princeton University in the United States, recently released an open source AI model called "Infinigen", which can generate realistic 3D natural world scenes.

It is reported that the Infinigen model is built on the open source modeling software Blender, which can generate a series of natural "plants, animals" scenes through random mathematical rules, and can customize the terrain by inputting prompt words, and add "clouds, rain, snow, thunderstorms , Skyfire” and other natural phenomena. However, the current efficiency of the Infinigen model needs to be improved. According to the test of the researchers, it takes 3.5 hours to generate a 1080P image (a pair of images of two scenes). In addition, the Infinigen model can also be used as a generator of training data, which can provide a series of features including "object detection", "semantic segmentation", "pose estimation", "3D reconstruction", "view synthesis" and "video generation". Scene training for computer vision-related AI.

LG releases Captioning AI, the first generative AI commercialization service

According to IT House's report on June 19, at the world's largest computer vision conference CVPR 2023 held in Vancouver, Canada yesterday, LG Artificial Intelligence Research Institute, a subsidiary of South Korean technology giant LG, disclosed the company's first generative AI commercialization service, Captioning AI, Provides a more efficient solution for customers who need to manage a large number of images.

According to LG, Captioning AI is based on zero-shot learning technology, which enables AI to use its previous accumulated experience and knowledge to explain the objects or scenes it sees for the first time, just like humans. On average, the service can generate 5 full descriptive sentences and 10 keywords for an image in 10 seconds. If the number of pictures is expanded to 10,000, these tasks will be completed within two days, so that a customized picture search and management system can be established in a short period of time. According to LG, Captioning AI was developed in-depth cooperation between LG Artificial Intelligence Research Institute and Shutterstock, one of the world's largest image galleries, which has extensive expertise in image labeling. In addition, the two parties also carried out AI ethics verification such as filtering and screening of learning data to ensure copyright transparency.

OpenAI CTO: Working on commercial subscription of ChatGPT

According to Fenghuang.com reported on June 16, OpenAI Chief Technology Officer (CTO) Mira Murati said that his team has begun to work on providing a commercial subscription to ChatGPT, which will allow customers to customize the ChatGPT version for specific purposes. They've also launched so-called plugins that allow users to stay inside ChatGPT while they're done using ChatGPT to browse the web, buy groceries, and make restaurant reservations through Instacart and OpenTable.

At present, OpenAI, like many software companies in Silicon Valley, bills itself as a "platform" company, offering two products: one is their application programming interface (API), which allows third parties to pay to integrate OpenAI's software into their products Medium; the other is ChatGPT. OpenAI's largest shareholder, Microsoft, will become its third independent revenue stream. Microsoft has invested tens of billions of dollars in OpenAI. Murati said, "ChatGPT allows us to reach people directly, gather feedback, and make our models more accurate and useful. The API is a platform that allows others to build on top of our models."

Mango Super Media: The virtual host "YAOYAO" has been used in many scenarios

According to a report by the Financial Associated Press on June 19, Mango Super Media stated on the interactive platform that the company's virtual host "YAOYAO" and intelligently guided virtual person "Zhen Cheng" have been used in multiple scenarios; the recently launched "All Staff Accelerating 2023" uses technologies such as AR and AI to allow guests to become digital human beings to participate in program task challenges, which effectively improves the content's highlights and fun points; in the future, the company will continue to explore the application of virtual human technology relying on its own unique content endowment new possibilities.

Meta has lost a third of researchers who published AI research results last year

According to the Financial Associated Press, citing the Wall Street Journal, analysis data shows that Meta has lost at least one-third of researchers who published AI research results last year. Many people expressed fatigue or lack of confidence in Meta and believed that Meta could not keep up with competitors. pace. A Meta spokesperson said the company has continued to recruit and bring in new AI talent.

Investment and Financing:

AI + cloud industrial software company "Sanweijia" completed the D round of hundreds of millions of yuan in financing, 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.

It is reported that Sanweijia was established in 2013. Starting from the vertical market segment of home furnishings in industrial software, it has successively launched three major product matrices: 3D cloud design, cloud manufacturing, and numerical control systems. "AI-based" and "integrated modeling/simulation/processing" cloud industrial software solutions. Previously, Sunway Home has successively received tens of millions of yuan in round A, 300 million yuan in round B strategic investment from Softbank China, GF Xinde, and Red Star Macalline, and 500 million yuan in round C strategic investment from Alibaba.

Ant and Baidu jointly invested in the AIGC track, and Shengshu Technology completed the first round of financing of nearly 100 million yuan, with a valuation of 100 million U.S. dollars

According to Titanium Media, on June 19, Beijing Shengshu Technology Co., Ltd. (hereinafter referred to as "Shengshu Technology") officially announced the completion of an angel round of financing of nearly 100 million yuan. This round was led by Ant Group and Baidu initiated BV Baidu. Venture capital and Zhuoyuan Capital followed suit, and the current valuation has reached 100 million US dollars.

This round of funds will be mainly used for the construction of the core R&D team to accelerate the development of multi-modal large models and application products. This is the first AIGC project invested by Ant Group since the release of ChatGPT in November last year. It is also the third AI content generation-related project invested by Baidu Venture Capital. The first two are Xihu Xinchen and Morph Studio .

It is reported that Shengshu Technology was established in March 2023. It was jointly incubated by Tsinghua-based AI company RealAI, Ant Group and BV Baidu Ventures. It is committed to becoming the world's leading builder of generative artificial intelligence infrastructure and applications. The company said that at this stage, Shengshu Technology is building vertical applications based on the underlying multi-modal general-purpose large model for subdivided scenarios of image creation and 3D asset creation. The AI image generation technology developed by it has exceeded the level of the latest version of the basic model of Stable Diffuion, and is expected to catch up with the latest version of Midjourney within this year.

Metaverse recommended reading:

"AI is crazily polluting the Chinese Internet"

The AI model obtains a large amount of Internet data, but cannot distinguish the authenticity and credibility of the information well. The result is that we have to deal with a flood of low-quality content that is generated rapidly, making people dizzy and dizzy. It's hard to imagine what the results will be like if the large models of ChatGPT are trained with this kind of data...

"China Literature Group CEO Hou Xiaonan issued an internal letter: embrace the historical opportunity of AIGC, establish four business divisions, and create a multi-modal IP experience and consumption platform"

On June 19, the internal letter exclusively obtained by Titanium Media APP from China Literature showed that Hou Xiaonan, CEO of China Literature, was launching a new round of organizational upgrading, and stated that the company’s medium and long-term business blueprint is to upgrade AIGC to empower original multiple A large platform of modal and multi-category content, building a new IP upstream and downstream integrated ecosystem.