The AI Tools Collection uses a monthly format of 3 issues, updating the latest AI industry news daily. This issue is the collection of AI industry news for May Issue 2 (May 11 to May 20).
May 20
- The National Supercomputing Center in Tianjin grandly released the "Tianhe Exascale Intelligent Computing Open Innovation Platform" and the "Tianhe Tianyuan Large Model," built on domestic Tianhe supercomputing power and intelligent computing power. (Read full article:Tianjin Government Affairs Network）
May 19
- Major news! The official ChatGPT iOS app is now available! In the early hours of today, OpenAI officially launched the ChatGPT iOS app on the App Store, but it is currently only available in the US App Store. (Store address:https://apps.apple.com/app/openai-chatgpt/id6448311069）
- According to foreign media reports and people familiar with the matter, Apple has restricted the use of ChatGPT and other external artificial intelligence (AI) tools, and the company is developing similar technology of its own. Documents show that Apple is concerned that employees using such AI programs may leak confidential data. Apple also told employees not to use Copilot on Microsoft's code hosting platform GitHub, which can automatically write software code. (Read full article:IT Home）
May 18
- StabilityAI launches the open-source version of the text-to-image tool DreamStudio StableStudio, aiming to collaborate with the broader community to create a first-class user interface for generative AI fully controlled by users. (Read full article:StabilityAI Blog）
- Google announced that Google Colaboratory (Colab) will add a brand-new AI coding feature, powered by the new "text-to-code" model Codey, supporting more than 20 coding languages, including Go, Google Standard SQL, Java, Javascript, Python, and Typescript, etc. (Read full article:IT Home）
- Users can now use Bing Chat without logging into a Microsoft account. However, there is a limit: users can have at most 5 conversation turns per topic, while logged-in users can reach 20. (Read full article:IT Home）
- Cloudwalk Technology released the "Congrong Large Model." In on-site demonstrations, the Congrong Large Model can support functions such as image-text understanding, copywriting, logical reasoning, multi-turn dialogue, programming, and factual question answering. (Read full article:Jiemian News）
May 17
- OpenAI CEO Sam Altman attended a hearing in the US Congress for the first time on the potential dangers of artificial intelligence technology, and Altman called for government regulation and intervention in generative artificial intelligence. (Read full article:QbitAI）
- Baidu's "Xiaodu" announced that it will launch the first product under the new brand "Xiaodu Qinghe" — the Xiaodu Qinghe learning phone, a learning phone designed specifically for teenagers. This learning phone will be equipped with a large AI learning model, enabling AI photo-based problem explanation, professional tutoring for Chinese and English AI essays, AI oral practice, and more. (Read full article:Cover News）
- Well-known meeting and office software Zoom invested in Anthropic (a strong competitor of OpenAI) and will integrate Anthropic's AI assistant Claude with Zoom. (Read full article:The Verge）
- iFlytek stated on its interactive platform that, in response to the common problems of large models, the Spark large model has a clear milestone plan for upgrades and iterations, saying that among domestic large model manufacturers, only this company has proposed a clear timetable for catching up with ChatGPT. (Read full article:IT Home）
May 16
- Kingsoft Office announced today,WPS AI it will embed large model (LLM) capabilities into four major components: Spreadsheets, Writer, Presentation, and PDF, supporting desktop computers and mobile devices. (Read the full article:ITHome）
- 360 and Zhipu AI have reached a strategic cooperation to jointly develop the hundred-billion-scale large language model 360GLM. 360 officially states that the model has reached the level of a new-generation cognitive intelligence general model. (Read the full article:ITHome）
- Xiaoice announced the launch of the "GPT Clone Plan", with the first batch limited to 300 spots. In as little as three minutes of data collection, an AI clone derived from one's own personality, skills, voice, and appearance can be created. (News source:36Kr）
May 15
- According to a message from the Midjourney AI WeChat official account, the official Chinese version of Midjourney has opened beta applications. It will be hosted on QQ Channels, uniformly opening at 6 PM every Monday and Friday, and the entry will be closed once a certain number of people is reached. (Read the full article:ITHome）
May 14
- According to The Korea Economic Daily, South Korea's two major tech giants — Samsung Electronics and Naver Corporation — have reached a cooperation agreement to jointly develop a generative AI platform for enterprise applications to compete with global AI tools such as ChatGPT. (Read the full article:ITHome）
- AI impersonated the famous British actor "Cumberbatch" Benedict Cumberbatch in a phone call to a small Spanish film company, saying he was interested in the screenplay and wanted to cooperate. But after several invitations to meet, Cumberbatch refused, and even proposed that the company first transfer him 200,000 pounds. (Read the full article:Cover News）
May 13
- Breaking! OpenAI tweeted that next week it will open internet access and numerous plugins to all ChatGPT Plus users! From Alpha to Beta, it will allow ChatGPT to access the internet and use more than 70 third-party plugins. (Read the full article:Wallstreetcn）
- The 14th Chinese Science Fiction Nebula Awards were announced in Guanghan Sanxingdui. Before the award ceremony, famous science fiction writer Liu Cixin accepted an interview with The Paper, discussing his views on the current development of artificial intelligence, saying, "The AI era will definitely come, but we should pay more attention to the present." (Interview video:The Paper）
- Domestic AI company Unisound announced that it will hold the Unisound Shanhai Large Model and Results Launch Conference on May 24, and will also release products and solutions for smart healthcare, smart IoT, and enterprise services based on the Shanhai large model. (Read the full article:ITHome）
May 12
- Known as ChatGPT's strongest rival, Claude has received an epic update, increasing the model's context window token count to a full 100,000, equivalent to 75,000 words! This means that even if you directly throw it materials of hundreds of pages and tens of thousands of words, such as financial reports, technical documents, or even a book, it can analyze and summarize them for you within a minute! (Read the full article:QbitAI）
- HuggingFace launched Transformers Agents, which can control more than 100,000 Hugging Face models to complete various multimodal tasks. (Read the full article:Machine Heart）
- At a recent earnings briefing, SoftBank Corp. CEO Junichi Miyakawa announced that the company established a new entity in March this year and recruited nearly 1,000 employees to participate in the research and development of "ChatGPT-like products." (Read the full article:ZAKER News）
- iFlytek released a record of investor relations activities. In response to controversy regarding the "iFlytek Spark large model," iFlytek replied: "The rumor that iFlytek Spark large model is 'a shell around OpenAI's ChatGPT' is inconsistent with the facts and also illogical. (Read the full article:ZAKER News）
- DLsite operator GEO Holdings announced a ban on uploading and selling illustrations generated using AI art tools such as Midjourney and Stable Diffusion. Following similar decisions by sites such as PixivFANBOX, this is the latest user-generated content platform to introduce such a ban. (Read the full article:IT Home）
May 11
- The Google I/O developer conference was held, where Google released and showcased a series of AI products and research, including the PaLM2 large model to counter GPT-4, Bard upgrades and the removal of the waiting list, Duet AI for Workspace, a search engine powered by generative AI, and the AI notebook project Project Tailwind. (Read the full article:ifanr）
- The operator of the large illustration site Fanbox, PixivFANBOX, announced a ban on uploading and selling illustrations generated using AI art tools such as Midjourney and Stable Diffusion. The official effective date and the definition of works generated by AI will be announced later. (Read the full article:IT Home）
- DeepMind co-founder and former Google vice president of AI products and policy Mustafa Suleyman recently warned during a speech at an investment forum in the California Bay Area that AI will seriously threaten the survival of the white-collar class in the next decade and bring about "a large number of losers." (Read the full article:IT Home）