Babbitt丨Daily Must Read: OpenAI Launches Customized Version of ChatGPT and GPT-4 Turbo Preview; xAI launches _IDE tool

Abstract: OpenAI held its first developer conference today and launched a service called “GPTs” for its ChatGPT, allowing users to create “their own version of ChatGPT” based on specific needs. OpenAI also launched a preview version of GPT-4 Turbo, a model that supports 128k contexts, is three times cheaper to input than GPT-4, and has double the rate limit.

Image source: Generated by Unbounded AI

Hot News:

OpenAI Launches User-Customized Version of ChatGPT and Launches GPT-4 Turbo Preview

According to a report by IT Home on November 7, OpenAI held its first developer conference today and launched a service called “GPTs” for its ChatGPT, allowing users to create “their own version of ChatGPT” according to specific needs. OpenAI says they introduced the features to give users “some control” over ChatGPT, such as businesses can create a special assistant that only employees can access, and parents can create a problem-solving tool that specializes in teaching their children how to solve tricky math problems.

OpenAI also launched a preview version of GPT-4 Turbo, a model that supports 128k contexts, three times cheaper input than GPT-4, double the rate limit, an updated knowledge base to April 2023, JSON Mode, and updated multiple function calling capabilities. GPT-4 is more powerful, cheaper, and cheaper to develop. In addition, OpenAI has introduced a new feature called JSON Mode, which makes it easier for developers to make it easier for models to call JSON and xml content, making it easier for models to return consistent output results for repeatable output, which can be useful for controlling model behavior and writing model unit test content.

Elon Musk’s AI company xAI launches IDE tool

Babbitt News, November 7, Elon Musk’s AI company xAI today announced the launch of xAI IDE, an integrated development environment for prompt engineering and explainability research. It accelerates prompt engineering through the SDK, which allows for the implementation of sophisticated hinting techniques and rich analytics capabilities to visualize network outputs.

According to xAI, the IDE can transparently provide transparent access to Grok-1, the model that powers Grog, to the engineers and researchers in the community. The IDE is designed to empower users to quickly explore the power of large language models. xAI also said it is now offering an IDE to members of the Grok Early Access Program.

OpenAI launches Assistants API, developers can customize AI assistants for their own applications

According to IT Home on November 7, at today’s first OpenAI developer conference, OpenAI launched a “specially built AI tool” Assistants API, which can use “additional knowledge” to help developers build AI assistants in their own applications.

According to OpenAI, the Assistans API provides three functions: “Code interpreter”, “Retri”, and “Function calling” to help users handle the “heavy lifting”. The Assistants API is available in beta today and can be experienced on the Assistants Playground home page without writing a single line of code.

In addition, OpenAI has launched the “Copyright Shield” feature to alleviate concerns about the potential infringement risk of generative AI. OpenAI says that if a user is required for legal claims for “copyright infringement” caused by AI-generated content, OpenAI will step in and defend the customer, and will also fully reimburse the costs incurred in the process. The functionality applies to ChatGPT Enterprise and all API users, but is exempted from liability if the user “has no intent to infringe” and is “cited from the source” under responsible AI use.

OpenAI: ChatGPT reaches 100 million weekly active users

According to a report by the Financial Associated Press on November 7, at OpenAI’s first developer conference, OpenAI CEO Sam Altman announced that ChatGPT currently has 100 million weekly active users. The service was launched nearly a year ago and gained about 100 million monthly active users within two months of its launch.

Microsoft partners with Inworld AI to bring AI-enhanced game development tools to Xbox

According to IT Home on November 7, Microsoft announced that it will work with Inworld AI to develop Xbox tools that enable developers to create AI-based characters, stories, and missions. According to reports, the cooperation will mainly include an “AI Design Copilot” system, with the help of which Xbox developers can create detailed scripts, dialogue trees, questlines, and more.

Haiyan Zhang, general manager of gaming AI at Xbox, said the partnership will bring together Inworld’s expertise in character development using generative AI models, Microsoft’s cloud-based AI solutions, including Azure OpenAI services, Microsoft Research’s technical insights into future games, and the Xbox team’s strengths in creator tools, to serve all developers.

Microsoft also mentioned that Inworld has been working on AI NPCs that can react to player actions on their own, much like ChatGPT or Bing Chat respond to user inquiries. They believe that these AI NPCs will have unique responses, such as tones, and bring to the game a storyline with a variety of possible dialogues or facets.

Meituan’s first AI chat product “Wow” was launched, which can talk to multiple AI characters with different personalities

On November 7, a “Wow” app developed by Shanghai Sankuai Worry-Free Purchase Technology Co., Ltd. has been launched on iOS and Android platforms. It is reported that the company is 100% controlled by Shanghai Hantao Information Consulting Co., Ltd., an affiliate of Meituan, which means that “Wow” is Meituan’s first AI chat product.

“Wow” is defined as a community of AI friends belonging to young people, and after logging in, users can talk to 29 AI characters with different characters, such as Socrates, Daji, Fusu, and psychological counselors, through voice or text. Asking the same questions for different AI roles will also have corresponding answers depending on the persona.

Source: Baidu has placed an order for AI chips with a total price of about 450 million yuan from Huawei, which is expected to be delivered by the end of the year

According to a Reuters report on November 7, two people familiar with the situation said that Baidu ordered AI chips from Huawei this year in response to the U.S. restrictions on the export of AI chips to China. Baidu has ordered 1,600 Huawei Technologies 910B Ascend AI chips for 200 servers, with a total order value of about 450 million yuan ($61.83 million), and Huawei will deliver all chips by the end of the year, the sources said.

Huawei has been working with Baidu since 2020 to make its AI platform compatible with Huawei hardware, according to its website. In August, the two companies said they would deepen the compatibility between Baidu’s Ernie AI model and Huawei’s Ascend chip.

Tongyi Qianwen Large Model App iOS version has been launched

Babbitt News, Alibaba Cloud’s Tongyi Qianwen App recently launched on the Apple App Store. According to the application introduction, Tongyi Qianwen is a super-large-scale pre-trained model, which can provide users with all-round assistance in many aspects such as creative copywriting, office assistants, learning assistants, and interesting life. Previously, the app was already available on major Android app markets.

Meta bans generative AI advertising tools in regulated industries

A Meta spokesperson said the company is banning advertisers from political campaigns and other regulated industries from using its new generative AI advertising product, as reported by Reuters on November 6.

Meta publicly disclosed the decision in an update released in its Help Center. “As we continue to test new generative AI ad creation tools in Ads Manager, advertisers running campaigns that align with housing, employment or credit or social issues, elections or politics, or health, pharmaceutical, or financial services are not currently permitted to use these generative AI features.” “We believe this approach allows us to better understand the potential risks and establish the right safeguards for the use of generative AI in advertising that involves potentially sensitive topics in regulated industries.”

Peking University and KLCII team proposed a training framework LLaMA-Rider, which allows large models to explore the open world autonomously

According to a report by Webmaster’s Home on November 7, a team from Peking University and Beijing Academy of Artificial Intelligence proposed a training framework called LLaMA-Rider, which aims to give large language models the ability to explore and learn tasks autonomously in the open world. This framework enables active exploration through a feedback-modification mechanism, allowing the model to receive feedback in the environment and continuously adjust its decision-making to gradually adapt to the open environment.

It is reported that LLaMA-Rider adopts a two-stage training method, first allowing the model to explore autonomously in the environment, and then integrating the successful experience into a supervised dataset for model fine-tuning and improving the ability to solve multiple tasks. Experimental results show that LLaMA-Rider has an advantage over reinforcement learning methods by showing high sampling efficiency and low training cost in multi-task solving. This study provides an important idea for the self-learning of large language models in the open world.

Investment & Financing:

German AI start-up Aleph Alpha receives over $500 million investment from Schwarz Group

German artificial intelligence startup Aleph Alpha has received more than $500 million in investment from Schwarz Group, owner of the Lidl supermarket chain, as reported by the Financial Times on November 6, and according to the statement, other investors include the venture capital arm of Bosch, Europe’s largest auto parts supplier. About a quarter of the funding came from equity investments, with the remainder made up of research grants and joint business development commitments that would not dilute shareholders’ equity, according to people familiar with the deal.

Founded in 2019, Aleph Alpha is one of Europe’s best-known start-ups focused on generative AI, whose software can mimic human abilities to create complex text, images, or computer code. The Schwarz Group, which is building a new IT department, is leading the deal with an artificial intelligence center in the southwestern German city of Heilbronn, which was set up by a foundation established by Lidl founder Dieter Schwarz.

Alpaca Education received 230 million yuan in strategic financing, which will be used to promote the company’s comprehensive AI strategy

According to a report by 36Kr on November 7, “Alpaca Education” recently announced that it has received $31 million in strategic financing, about 230 million yuan, and the investor in this round of financing is Hunan Incremental Capital. This round of financing will be used to promote the company’s comprehensive AI strategy, and further iterate the company’s research and development of the company’s vertical model, intelligent customer service, digital human teacher, etc. Among them, the company’s vertical model is a professional language learning model developed by Alpaca Education for the language training industry; Intelligent customer service is an intelligent Q&A system created by Alpaca Education using AI technology, which can quickly and accurately answer students’ questions. The digital human instructor is a virtual teacher created by Alpaca Education using virtual reality and AI technology, which can provide students with a more vivid and vivid teaching experience.

AIGC Recommended Reading:

“OpenAI’s first AI Spring Festival Gala, startup slaughter night!” GPT-4 Explosion Update, API Jumping Price Sale

Altman, CEO of OpenAI, shared dozens of new features and improvements to GPT-4 and reduced pricing for many of the platform’s services: the new GPT-4 Turbo model, which is more powerful, cheaper, and supports 128K contextual windows. The most important thing is that the GPTs function has been released, which allows each user to make their own “customized ChatGPT”, and can also make money for themselves through the upcoming “GPT Store”!

“The development tools behind Musk’s version of ChatGPT are online! xAI products are launched twice in a row, netizens: the delivery speed is too crazy”

Musk’s version of ChatGPT has just attracted a wave of eyeballs, and xAI’s second large-scale model product has suddenly appeared! Just now, Musk’s xAI officially announced: the launch of IDE, an integrated development environment for prompt engineering and explainability research.

When OpenAI wants to “take it all”, what opportunities are there for entrepreneurs? |Investors Say》**

Alpha Commune: OpenAI released a series of “explosive” updates at DevDay held in the early hours of this morning, also showing its “ambition” to “take it all”. In the face of OpenAI, Google, Microsoft, Meta, Baidu, Alibaba and other leading companies in AI capabilities and gradually improving product ecology, what opportunities do entrepreneurs have?