ChatGPT for Android is online; Miaoya camera responds to AIGC event daily suspected of infringing users’ rights and interests.
07 / 26
Highlights of global AIGC industry
1. The research on watermarking large model won the ICML 2023 Outstanding Paper Award.
2. The AI classifier tool has been disabled by OpenAI.
3. OpenAI is developing an open source model, code-named G3PO.
4. Microsoft CEO：Azure OpenAI has 11,000 customers.
5. The AI summary function of Microsoft Store has launched a preview version.
6. Microsoft, Google and OpenAI jointly set up the Frontier Model Forum.
7. The open material counting alliance was established and initiated by Shenzhen Stock Exchange and Tencent Cloud.
8. Meta extended Llama function can perform multilingual speech recognition.
9. Android 14 will include generative AI tools.
10. The Android version of ChatGPT was officially launched.
11. The open source traffic model of Beijing Jiaotong University can be commercialized for free.
12. Tsinghua undergraduates create a free version of desktop Copilot.
13. Chongqing AI Industry Three-year Action Plan was released.
14. Shanghai Consumer Protection Committee: Miaoya camera is suspected of infringing on fair trade rights.
15. Miaoya camera responds to the dispute. AI can be refunded if it fails to generate the avatar.
16. Netease has officially released the "Zi Yue" big model.
17. Nail 12 product lines and connect more than 40 scenes to the big model.
18. Runway AI video generation tools are fully open.
19. CAMBRIAN jointly won the bid for 753 million yuan intellectual calculation project.
20. Hengshuo’s integrated AI chip is expected to be rolled out in the third quarter.
21. IDC: By 2026, 50% of terminals will be equipped with hardware AI engine function.
22. Guanglianda: A large model layer of AI in the construction industry has been built.
23. UFIDA applied for a large model trademark.
24. Wayfair launched the AI Virtual Room Redesign.
25. GitHub, Hugging Face and others called on the EU to protect open source innovation.
26. The US Senate passed an amendment to require notification of technology investment in China.
27. The two parties in the United States warned about the risk of AI: it may be used to develop biological weapons.
28. Foreign media speculated that OpenAI was about to open source GPT-3.5.
29. The annual salary of the chief engineer of OpenAI is 6.6 million, ranking first in the world.
30. Alibaba Cloud supports Llama2 full range of training deployment.
31. Little Red Book Online AI Wensheng Map New Function "Now"
32. Meitu Xiu Xiu launched AI removal function.
33. The local question-and-answer model localGPT was launched.
34. Intel launched 34 open source AI reference kits.
35. AI is affecting the $200 billion game industry.
36. Synthetic data company Nimbus Intelligent completed angel+round financing.
37. The code model CodeGeeX2-6B is open source.
38. Ministry of Industry and Information Technology: Strengthen the forward-looking layout of future industries such as general artificial intelligence
1. The research on watermarking large model won the ICML 2023 Outstanding Paper Award.
ICML 2023, one of the international AI academic summits, was held at the Hawaii Convention Center from July 23rd to 29th. Today, ICML officially announced six outstanding thesis awards, including a watermark for large language models by the University of Maryland.
In this paper, a method of adding watermark to the output of large-scale language model is proposed, which embeds signals into the generated text. These signals are invisible to human beings, but can be detected by algorithms. Watermarks can be generated without retraining the language model and can be detected without accessing API or parameters. This paper also puts forward a statistical test method with interpretable P value and an information theory framework for analyzing watermark sensitivity.
Paper address:
https://openreview.net/forum? id=aX8ig9X2a7
2. The AI classifier tool has been disabled by OpenAI.
In an updated blog post, OpenAI said that it decided to stop using the AI classifier tool to distinguish between human-written texts and AI-generated texts from July 20th, because the tool has low accuracy and is not completely reliable. OpenAI is currently studying more effective text source technology, and has promised to develop and deploy mechanisms to enable users to understand whether audio or visual content is generated by AI.
3. OpenAI is developing an open source model, code-named G3PO.
According to The Information on July 24th, the open source large-scale language model being developed by OpenAI, currently code-named G3PO, has not yet decided on the release schedule.
4. Microsoft CEO：Azure OpenAI has 11,000 customers.
According to reports, in the conference call after Microsoft released the fourth quarter financial report of fiscal year 2023, Microsoft CEO Satya Nadella said that the sales of Azure cloud computing services accounted for more than half of its $110 billion cloud computing revenue in fiscal year 2023. This is the first time that Azure has accounted for most of Microsoft’s annual cloud computing business. Part of the growth was driven by Azure OpenAI. Azure OpenAI currently has 11,000 customers, a significant increase from the 4,500 revealed in mid-May. Microsoft intelligent cloud’s revenue in the fourth quarter was $23.993 billion, and it is estimated that the revenue in the first quarter of fiscal year 2024 will be $23.3 billion to $23.6 billion.
5. The AI summary function of Microsoft Store has launched a preview version.
According to the report of IT House on July 25th, the AI summary function of Microsoft Store has launched a preview version in the United States. This function can summarize thousands of online comments of an application into a concise text in a few seconds, highlighting the most important points. At present, these functions are only available to users of Windows beta in the United States, and the Microsoft Store application needs to be updated to version 22306.1401.x.x.
6. Microsoft, Google and OpenAI jointly set up the Frontier Model Forum.
Google, Microsoft, OpenAI and Anthropic issued a joint announcement today, announcing the establishment of the Frontier Model Forum, an industry organization dedicated to the safe and responsible development of cutting-edge artificial intelligence models. The Frontier Model Forum will establish an advisory committee to help guide its strategy and priorities.
According to the announcement, the core objectives of the forum are:
(1) Promote the research of artificial intelligence security, promote the responsible development of cutting-edge models, minimize risks, and realize independent and standardized assessment of capabilities and security.
(2) Determine the best practices of responsible development and deployment of frontier models to help the public understand the nature, capabilities, limitations and impacts of technology.
(3) Cooperate with policy makers, scholars, civil society and enterprises to share knowledge about trust and security risks.
7. The open material counting alliance was established and initiated by Shenzhen Stock Exchange and Tencent Cloud.
Cailian on July 26th, Cailian reporter learned from Shenzhen Data Exchange that Shenzhen Data Exchange and nearly 50 units set up an "open material counting alliance". Co-sponsors include Tencent Cloud, Huada Gene, Yuntian Lifei, Obi Zhongguang, Ubisoft, China Unicom Guangdong Branch, Huaao Data and other enterprises, associations such as Shenzhen Artificial Intelligence Society, think tanks and research institutions such as National Supercomputing Shenzhen Center, and universities such as Shenzhen Graduate School of Peking University, Harbin Institute of Technology (Shenzhen) and Hong Kong Institute of Technology. It is reported that the alliance will focus on high-quality Chinese training data and multimodal training data, coordinate the formulation of data elements, data governance, training data, data labeling, synthetic data and other related standards, and assist data exchanges to add new categories and new areas related to large models.
8. Meta extended Llama function can perform multilingual speech recognition.
The Hugging Face page shows that Meta AI published a new paper on July 21st, expanding the functions of its large language model Llama. The R&D team extends the function of the large language model by directly attaching a small audio encoder to enable it to perform speech recognition. By adding a series of audio embedments directly before embedding text tags, large-scale language models can be transformed into automatic speech recognition (ASR) systems and used in exactly the same way as their text counterparts. Multilingual LibriSpeech(MLS) experiment shows that although LLMA is mainly trained on English texts, the performance of Llama-7B can be improved by 18% compared with the monolingual baseline by integrating the consistency encoder, and it can perform multilingual speech recognition.
Paper address:
https://arxiv.org/abs/2307.11795
9. Android 14 will include generative AI tools.
Cailian, July 26th, sundar pichai, CEO of Alphabet, said that Android 14 will include generative AI tools, and 80% of advertisers will use at least one AI-driven search product.
10. The Android version of ChatGPT was officially launched.
OpenAI announced in a tweet last night that the Android version of ChatGPT can now be downloaded in the United States, India, Bangladesh and Brazil. OpenAI plans to extend it to more countries and regions next week.
Portal:
https://play.google.com/store/apps/details? id=com.openai.chatgpt&pli=1
11. The open source traffic model of Beijing Jiaotong University can be commercialized for free.
Recently, Beijing Jiaotong University and other institutions have opened up a general common sense transportation model TransGPT with 7 billion parameters, which is not only completely open to academic research, but also can be commercialized for free after applying by mail and obtaining official commercial license.
According to reports, TransGPT is the first open source transportation model in China, which can realize the functions of traffic situation prediction, intelligent consulting assistant, public transportation service, traffic planning and design, traffic safety education, assistance management, traffic accident report and analysis, automatic driving assistance system, etc. It can provide general knowledge for road engineering, bridge engineering, tunnel engineering, road transportation, waterway transportation, urban public transportation, transportation economy, transportation safety and other industries.
Code address:
https://github.com/DUOMO/TransGPT
12. Tsinghua undergraduates create a free version of desktop Copilot.
According to Machine Heart’s report on July 25th, several undergraduates from Tsinghua University Computer Science Department joined in generative AI entrepreneurship after graduation, and launched Copilot desktop intelligent assistant "Ling Yu Assistant" which can be used with any software. Whether using PDF reader, browser, WeChat or VSCode, users can call ChatGPT capability anytime and anywhere. The AI assistant currently supports Windows and macOS downloads. The current version is built on GPT-3.5 and is open to all users free of charge. The paid version has a professional version of $9 per month and an elite version of $19 per month, and does not support image input for the time being. The team said that GPT-4 and Claude will be introduced in the future.
13. Chongqing AI Industry Three-year Action Plan was released.
On July 25th, official website, the Economic and Information Commission of Chongqing, publicly issued a notice on printing and distributing the Action Plan for High-quality Development of Artificial Intelligence Industry Driven by Scenes in Chongqing (2023-2025). According to the action plan, its development goal is to successfully create a national pilot area for artificial intelligence innovation and application, build 10 benchmark scenario projects, form 100 artificial intelligence products with core competitiveness in key areas and typical scenarios, cultivate 10 leading enterprises of artificial intelligence above 100 million level, gather 100 enterprises above artificial intelligence scale and 1,000 artificial intelligence related enterprises, and build 3-5 artificial intelligence industrial clusters.
The key tasks pointed out in the document include optimizing and expanding the supply of artificial intelligence products, making efforts to break through artificial intelligence software, implementing artificial intelligence software to "unveil the list", and promoting the research and development of independent artificial intelligence framework, operators and other key technologies such as AIGC(AI Generated Content); Accelerate the cultivation of artificial intelligence hardware, encourage the development and scene application of neural network processor chips, image processing chips and voice processing chips, and promote the development of high-end intelligent chips.
14. Shanghai Consumer Protection Committee: Miaoya camera is suspected of infringing on fair trade rights.
WeChat official account, WeChat of Shanghai Consumer Protection Committee, issued a document today, pointing out that the recently popular AI photo applet Miaoya Camera marked "Once purchased successfully, no refund is supported" in small gray letters on the payment page, which is suspected of infringing consumers’ fair trading rights. There is no refund related clause in its service agreement. In addition, the phone number left by the registered company of this applet for Wuxu Network Technology (Shanghai) Co., Ltd. is either unanswered or does not exist.
15. Miaoya camera responds to the dispute. AI can be refunded if it fails to generate the avatar.
Today, Miao Duck Camera, a popular AI photo applet, made three responses to the discussion triggered by the service: "1. Miao Duck Camera provides a service based on computing power, so the service is considered complete after the photo is generated. We solemnly promise that if the digital avatar generation fails due to technical problems, we can get a refund and generate it again for free. 2. AI algorithm has certain uncertainty, and accessories such as glasses may be unrecognizable. To this end, we will optimize the algorithm and solve the problem as soon as possible. Users who encounter similar problems can consult customer service. 3. Because downloading photos involves virtual currency recharge, virtual currency and cash can only flow in one direction at present, and the recharge agreement has been clearly stipulated, so no refund is supported at present. "
Miaoya camera was officially opened on July 1 this year, and the relevant team of Ali Youku Tudou was behind it. Users can upload 20 photos, pay 9.9 yuan, and then they can make exclusive ID photos, workplace photos, photo photos or style blockbusters. The synthetic style is similar to the works of photography institutions such as Tianzhenlan and Haima.
▲ Miaoya camera responds to controversy
16. Netease has officially released the "Zi Yue" big model.
Today, according to 36Kr, Netease has launched the first vertical model of education in China, and released six innovative applications based on the model-LLM Translation, Virtual Human Oral Coach, AI Composition Guidance, Grammar Exquisite, AI Box and Document Question and Answer.
17. Nail 12 product lines and connect more than 40 scenes to the big model.
Nail WeChat WeChat official account issued a document on July 25th, announcing that 12 product lines and more than 40 scenes have been connected to the big model. The nail slash "/"now includes the functions of creating an exclusive question-answering robot, real-time conference voice to text, in-session summary, dialogue creation project, intelligent recommendation project template, intelligent email writing, intelligent document, dialogue insertion chart, dialogue generation PPT, and automatic creation of mind map framework.
18. Runway AI video generation tools are fully open.
On July 24th, Runway, an AI video generation company, tweeted that the restrictions on Gen-1 and Gen-2 video generation were lifted. You can try to generate videos for free by registering your account. Each generated video lasts for 4 seconds and consumes 5 points per second. The free version has 125 points, which can generate 8-second video with Gen-1, 25-second video with Gen-2 or 25 images. If the free points are exhausted, the payment standard is $0.01/point.
19. CAMBRIAN jointly won the bid for 753 million yuan intellectual calculation project.
CAMBRIAN disclosed the announcement of winning the bid on July 25th, and announced that CAMBRIAN formed a consortium with Taizhou Branch of China Mobile Communications Group Zhejiang Co., Ltd. and Zhejiang Public Information Industry Co., Ltd., and won the bid for the project of "Digital Infrastructure Upgrading Project of Southeast Zhejiang Digital Economy Industrial Park (Phase I)", with an estimated bid amount of 753 million yuan. According to the division of labor of the member units of the consortium, CAMBRIAN, as the leader of the consortium, is responsible for the supply, installation and follow-up service of the intelligent computing hardware. It is estimated that the company’s share will be about 70% of the total price of the project, and the actual amount will be subject to the formal contract finally signed.
20. Hengshuo’s integrated AI chip is expected to be rolled out in the third quarter.
Cailian, July 26th, Hengshuo, a storage chip and MCU chip company, said on the interactive platform that the company’s integrated AI chip is expected to be streamed in the third quarter.
21. IDC: By 2026, 50% of terminals will be equipped with hardware AI engine function.
IDC consulted WeChat WeChat official account on July 25th, and predicted that by 2026, nearly 50% of terminal devices in China market will have AI engine technology. IDC believes that in the future, AI terminals will be widely developed in the following ten fields: creation and creativity, games and virtual worlds, speech synthesis and conversion, vision and image processing, chat bots, personalized recommendation, medical assistance, and face recognition will be widely used in the terminal field & computers will automatically classify pictures, automatically generate content, and terminals will become more secure.
22. Guanglianda: A large model layer of AI in the construction industry has been built.
According to the interface news report, Guanglianda, an Internet platform service company in the construction industry, said on the investor interaction platform on July 25th that AI is one of the important underlying technologies to support digital architecture. As early as 2015, the company set up an AI team to conduct research on building big data and AI-related technologies. At present, it has built a three-tier AI technology system in the construction industry, including AI big model layer, tool platform layer and product application layer.
23. UFIDA applied for a large model trademark.
According to Tianyancha App, UFIDA Network Technology Co., Ltd. recently applied for the trademark of "UFIDA Enterprise Service Model", which is internationally classified as advertising sales and the trademark status is under application. UFIDA has previously applied for related trademarks such as YONGPT and YONYOUGPT, and the international classification involves scientific instruments, website services, advertising sales, etc., all of which are under application. UFIDA was established in 1995, and its legal representative is Wang Wenjing. Its business scope includes technical development of computer software, hardware and external equipment, enterprise management consulting, database service, and sales of computer software and hardware and external equipment.
24. Wayfair launched the AI Virtual Room Redesign.
According to reports, Wayfair, the mainstream home e-commerce platform in the United States, has launched a free virtual room redesigner Decorify. When the user uploads photos of the room, selects the desired visual style, and inputs the type of room to be decorated (currently only the living room is supported), it uses generative AI to show the redecorated version of the room that the user wants. Users can select a single item in the room and let Wayfair’s AI model replace it with a certain style item in the drop-down menu. The website can be run on a computer or a mobile phone.
Portal:
https://www.wayfairnext.com/decorify
25. GitHub, Hugging Face and others called on the EU to protect open source innovation.
According to the webmaster’s home, an alliance of open source AI stakeholders such as Hugging Face, GitHub, EleutherAI, Creative Commons, LAION and Open Future called on EU policy makers to protect open source innovation when formulating the EU AI Bill. In their published policy documents, they emphasized the importance of open source AI, pointed out that supervision should not hinder the innovation of open source AI, and suggested that the discussion of the AI bill should provide clearer information for open source developers to avoid adverse effects on the open source AI ecosystem. They also hope that through the influence of the European Union, they can lead the global technical supervision dialogue and let open source developers play a greater role in the policy-making process.
26. The US Senate passed an amendment to require notification of technology investment in China.
According to Reuters, quoted by the World Wide Web, on July 25th, the US Senate passed an amendment to the "National Defense Authorization Act for FY 2024" with an overwhelming majority, which will require American enterprises to inform federal agencies of their investments in China’s technology fields (such as semiconductors and AI).
27. The two parties in the United States warned about the risk of AI: it may be used to develop biological weapons.
According to reports, on Tuesday, local time, both US Democratic and Republican senators warned about the possibility of AI being maliciously exploited, focusing on the possibility of AI being used to create biological attacks.
At the hearing of a Subcommittee of the Senate Judiciary Committee, Dario Amodei, CEO of Anthropic, a generative AI unicorn company, said that AI can help other unskilled malicious actors develop biological weapons. He said: "We believe that by enabling more actors to carry out large-scale biological attacks, this poses a serious threat to US national security."
28. Foreign media speculated that OpenAI was about to open source GPT-3.5.
According to foreign media Analytics India Mag, Andrej Karpathy, a well-known figure in the field of deep learning in OpenAI, recently said in Twitter that all these are very common for the Transformer language model in the Llama craze. "If/when OpenAI publishes the model as a weight (I can neither confirm nor deny it! ), then most of the code here will be very relevant. The foreign media reporter believes that according to Karpathy, OpenAI is likely to open source GPT-3.5. This conversation originated from a post on Twitter, in which a user asked Karpathy why he had been playing Llama 2 instead of building Jarvis for OpenAI.
29. The annual salary of the chief engineer of OpenAI is 6.6 million, ranking first in the world.
According to CSDN, the data collection website Levels.fyi recently released the global programmer income report for the first half of 2023. According to the report, the median total salary of the chief engineers on the list is above $660,000, which is a significant increase compared with last year. Among them, OpenAI ranked first with a median total salary of 925,000 US dollars, and last year it was just a "nobody". This daily salary is mainly for software engineers with 10 years of working experience, and its salary structure includes: a basic salary of 300,000 US dollars+625,000 "equity". Among them, "equity" is not paid every year, but employees receive it every four years.
30. Alibaba Cloud supports Llama2 full range of training deployment.
According to Alibaba Cloud’s official WeChat official account, Alibaba Cloud is the first in China to launch a training and deployment plan for the full range of Llama2 versions, and all developers are welcome to come to Alibaba Cloud to build exclusive large models. For the convenience of developers, PAI, a machine learning platform in Alibaba Cloud, took the lead in deeply adapting Llama2 series models in China, and put forward practical schemes of light-weight fine-tuning, full-parameter fine-tuning and reasoning service.
Developers can click "Notebook Rapid Development" on the Llama2 model page of the magic community, and pull up the Alibaba Cloud machine learning platform PAI with one click to develop and deploy the model on the cloud; The Llama2 model downloaded from other platforms can also be developed by PAI in Alibaba Cloud. It is understood that the Llama2 language model has recently announced open source, covering 7 billion, 13 billion and 70 billion parameter versions.
31. Little Red Book Online AI Wensheng Map New Function "Now"
Xiaohongshu recently launched an AIGC function called "Now". From the creative entrance "+"to "now", users can input words, and the corresponding pictures are generated by the algorithm according to the words, including notepad, chat sentiment, spit and other styles. If you don’t like the automatically generated picture, you can click "Change it for a new one".
32. Meitu Xiu Xiu launched AI removal function.
According to the official WeChat official account of Meitu Xiu Xiu, Meitu Xiu Xiu recently launched the AI removal function. Users only need to "frame" or "smear" to intelligently remove unwanted elements in photos, and they can also intelligently fill the background and complement the surrounding elements, thus achieving a convenient, seamless and natural removal effect. When there is an unnatural situation after removal, the user can continue to check the removal area, and AI will continue to correct and optimize until the user is satisfied.
33. The local question-and-answer model localGPT was launched.
According to the webmaster’s home, the local question-and-answer model localGPT was recently launched on the code hosting platform GitHub. LocalGPT is a completely offline question-and-answer tool, which aims to protect users’ privacy. It is built with LangChain and Vicuna-7B models, and it can ask and answer documents based on local data without networking. Users can import text documents (such as txt, pdf, csv and other formats) into the tool, and get accurate answers by asking questions, and all data will not leave the user’s environment. The tool is still in the testing stage. Vicuna-7B model is based on Llama model and has the original Llama license.
Project address:
https://github.com/PromtEngineer/localGPT
34. Intel launched 34 open source AI reference kits.
On July 24th, Intel announced the new progress of its one-year cooperation with Accenture, and now it provides 34 open source AI reference tool kits to the community. Each tool suite includes model code, training data, machine learning pipeline instructions, libraries and oneAPI components to optimize artificial intelligence and make it accessible to organizations in multi-architecture local, cloud and edge environments.
35. AI is affecting the $200 billion game industry.
It is reported that AI has begun to influence the game industry worth 200 billion dollars. The head of a large studio in Japan predicted that half of its programmers and designers would be eliminated within five years. Industry analyst Serkan Toto believes: "AI may eventually eliminate the whole job category in the game, such as quality control, debugging, customer support or translation."
Kenji Fukuyama, an analyst at UBS Securities, said that AI can halve the capital and time investment in large-scale game projects. Preferred Networks, Japan’s largest AI startup, has cooperated with Crypko, an animation character generation platform. Usually, the outsourcing cost of each character illustration is as high as 100,000 yen (about 5,058 RMB), while the monthly fixed cost of each image obtained from Crypko is 4,980 yen (about 252 RMB) and the commercial license fee is 980 yen (about 50 RMB).
Gala Sports, a China-listed head sports game manufacturer, asked department heads to study machine learning and offered a $7,000 prize for novel AI ideas. Its CEO Jia Xiaodong believes: "The impact of AI on the game industry in the past three or four months may be as great as the changes in the past thirty or forty years." Wangchen Technology uses open AI services to build an internal toolkit for rendering realistic 3D head models, thus greatly reducing the task cost of two weeks and up to 200,000 yuan in previous outsourcing, and now it only takes half a day.
36. Synthetic data company Nimbus Intelligent completed angel+round financing.
Science and technology innovation board Daily reported on the 25th that synthetic data company "Nimbus Intelligence" recently completed Angel+round financing. The company has completed seed round, angel round and angel+three rounds of financing, with the accumulated financing amount reaching tens of millions of RMB. The investors include SEE Fund, Qiji Chuangtan, Chentao Capital and variable capital.
Founded in January 2023, Nimbus Intelligence is located in Beijing, and is committed to providing synthetic data solutions for enterprises to land AI, and combining generative AI and simulation technology to provide 3D, physically realistic and generalizable synthetic data for the industry.
37. The code model CodeGeeX2-6B is open source.
Yesterday, the GLM technical team announced in the official WeChat official account that the code generation model CodeGeeX2-6B was open source. At the same time, GLM technical team will continue to iteratively upgrade the model to provide more powerful code assistance.
CodeGeeX2 is the second generation model of multi-language code generation model CodeGeeX, which is implemented by injecting code based on ChatGLM2 architecture. Compared with StarCoder-15B with 15 billion parameters, CodeGeeX2-6B has an advantage of nearly 10% with 6 billion parameters. The following are the evaluation results of CodeGeeX2 on HumanEval.
Installation address of CodeGeeX2-6B:
https://github.com/THUDM/CodeGeeX2
38. Ministry of Industry and Information Technology: Strengthen the forward-looking layout of future industries such as general artificial intelligence
According to Cailian, the symposium of responsible comrades of the national industrial and information technology authorities was held in Beijing on July 25th. The meeting emphasized that the key industrial chains should be effectively guaranteed to be autonomous and controllable, the overall scheduling and supervision should be strengthened, the responsibilities of all parties should be compacted, industrial policies should be improved, the construction of talent teams should be strengthened, and the quality of industrial development and the advantages of the whole industrial chain should be enhanced. Efforts will be made to cultivate and expand emerging industries, do a good job in the construction of national manufacturing innovation centers, enhance industrial scientific and technological innovation capabilities, strengthen the forward-looking layout of future industries such as general artificial intelligence, 6G and quantum technology, and create new competitive advantages.
Original title: "ChatGPT for Android is online; Miaoya camera responded to suspected infringement of user rights; Netease has a way to publish the big model AIGC Daily.
Read the original text