Along with various AI models competing to bloom, there is also a crazy influx of capital; As major companies increase the protection of their products, transparency is declining; The powerful ability and wide application of AI have also attracted the attention of policy makers in various countries, and have introduced relevant policies to protect its development. Image source: Physicist Organization Network
[World tide of scientific and technological innovation]
◎ reporter Liu Xia
A heavyweight product in the field of artificial intelligence, "You sing and I debut". Open AI GPT-4 ushered in the "first show"; Google launched a test version of the chat robot "Bard". Before that, the company had already launched the medical large-scale language model Med-PaLM, as well as the large-scale language models PaLM API and MakerSuite；. Anthropic, an artificial intelligence startup, also released Claude；, a product similar to ChatGPT, on the 15th. Coincidentally, China Tsinghua University launched ChatGLM, a bilingual dialogue robot.
AI technology ushered in the spring of "letting a hundred flowers blossom". In the report on the 22nd, Physicist Organization Network listed some trends worthy of attention in the AI field: with various AI models competing to bloom, there is also a crazy influx of capital; As major companies increase the protection of their products, transparency is declining; The powerful ability and wide application of AI have also attracted the attention of policy makers in various countries, and have introduced relevant policies to protect its development.
Deployment is on the rise.
In November last year, ChatGPT was born, and AI once again took the center of the technology stage.
Just two months later, ChatGPT has more than 100 million monthly active users, making it the fastest growing consumer application in history. Earlier this month, OpenAI released the ChatGPT API, which will lower the threshold for developers to integrate ChatGPT into their own applications and services.
OpenAI emphasized that some companies including fresh e-commerce platform Instacart, cross-border e-commerce platform Shopify, photo sharing application Snap, word recitation application Quizlet, etc. have taken the lead in accessing ChatGPT API to improve the efficiency and experience of customer service, marketing and education.
In the past week, ChatGPT and other technologies have further accelerated the pace of "attacking the city": at present, the American software service company Notion, the social question and answer website Quora and the search engine DuckDuckGo have integrated Claude into their products. The first customers of the latest GPT-4 released by OpenAI also include Duolingo, a language learning application, Morgan Stanley, etc. The Icelandic government is also using GPT-4 to protect its language.
Competition is becoming increasingly fierce.
As AI’s capabilities become stronger and wider, the transparency of many companies becomes lower and lower.
The most obvious thing is OpenAI’s technical report on GPT-4, which writes: "In view of the fierce competition environment faced by large-scale models such as GPT-4 and security considerations, the report does not contain further details on architecture (including model size), hardware, data set construction, training methods, etc."
Still, some companies have released more details. For example, Meta publicly released LLaMA, and put forward licensing requirements for commercial use. Together has launched OpenChatKit 0.15, a chat robot. Developers can fully access its source code and the data set used to train the robot. ChatGLM-6B in Tsinghua University is also open source.
A large influx of funds
AI technology and chat bots have attracted a lot of money, the most notable of which is that Microsoft has thrown $10 billion for OpenAI.
According to a report by Forbes magazine on the 15th, American startup Adept AI Labs has raised $350 million in Series B financing, with a post-investment valuation of at least $1 billion. In February this year, Google invested $300 million in Anthropics. In addition, Character AI raised $200 million from Anderson Holovitz Venture Capital this month; Perplexity AI, established in 2022, completed a financing of US$ 25 million led by Enyi Investment Company. After financing, the company’s valuation will reach US$ 150 million.
Salesforce Venture Capital also launched a $250 million fund and invested in a series of AI startups including Anthropi, Cohere and You.com.
Multi-country development blueprint
These AI models have become "frequent visitors" of many media and attracted the attention of decision makers all over the world.
In October 2022, the White House released the blueprint of the bill of rights for artificial intelligence. Recently, Alexander Madrid, a professor at the Massachusetts Institute of Technology, proposed to the Committee on Cyber Security, Information Technology and Government Innovation of the US House of Representatives: "Artificial intelligence is no longer a science fiction novel, nor is it a technology limited to the laboratory. Artificial intelligence is a technology that has been widely deployed and adopted, and it will completely change our lives; We now need to think about how to shape the world driven by artificial intelligence. "
In the past week, Britain announced plans to set up a basic model task force reporting directly to the Prime Minister. At the same time, the British government also announced that it will invest 900 million pounds to build a billion-dollar second-class cutting-edge supercomputer, and will also develop its own chat robot "BritGPT". As British Prime Minister Sunak said: "Basic models like ChatGPT are beginning to show extraordinary new capabilities … … This is the new reality we are facing! The competition to create, develop and utilize these new technologies has flourished around the world. "