Recently, at the Microsoft Inspire conference, Microsoft announced the latest progress in the commercialization of its GPT-4 application based on OpenAI, launching a new artificial intelligence subscription service that will charge Office 365 E3, E5, Business Standard and Business Premium subscription merchants $30 per person per month.
The release of new AI services and Microsoft's market value increased by trillions, hitting a record high
Currently, Microsoft 365 Business Standard and Business Premium users cost $12.5 and $22/month, respectively, and E3 and E5 $36 and $57 per month, respectively. For E3 users, adding Copilot means an 83% increase in fees; For Business Standard users, the cost almost triples. Affected by this news, the capital market has been chasing all the way, and Microsoft's market value once rose by 154 billion US dollars (about 1.1 trillion yuan). The market capitalization reached a record high of $2.67 trillion.
According to Microsoft's official website, Microsoft 365 Copilot can provide ideas in Word, analyze data in Excel, design presentations in PPT, classify Outlook inboxes, make meeting minutes in Teams meeting software and so on. On the one hand, many white-collar jobs will become easier, but on the other hand, many white-collar workers may be at risk of losing their jobs: after all, for as little as $30 a month, you can do creative director, data analyst, administration, marketing and other jobs, which is not too much for bosses?
Multi-service launched, AIGC heating up
At the same time, Microsoft launched a preview version of Bing Chat Enterprise, which means that more than 160 million people can already use Bing Chat for $5/month. Microsoft also launched visual search in Bing Chat, allowing users to ask questions based on entered pictures to get answers. What's more, at Microsoft's AI infrastructure service Azure OpenAI Service, Meta today announced that LLaMA 2, a free and open source large-language model, was also launched. This means that Azure customers will be able to easily and securely fine-tune and deploy Llama 2 models with 7 billion, 13 billion, and 70 billion parameters on Azure, quickly building dedicated large models for their own business. The launch of these new tools means that the race between tech giants such as Microsoft, Google, and IBM in the field of generative artificial intelligence tools is heating up.
The wave of commercial use of large models in mainland China is about to begin
Recently, the relevant person in charge of the Cyberspace Administration of China said that the Cyberspace Administration of China, together with the National Development and Reform Commission, the Ministry of Education, the Ministry of Science and Technology, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Administration of Radio, Film and Television, promulgated the Interim Measures for the Management of Generative Artificial Intelligence Services, which will come into force on August 15, 2023.
The Measures put forward that the state adheres to the principle of attaching equal importance to development and security, promoting innovation and governing according to law, adopts effective measures to encourage the innovative development of generative artificial intelligence, implements inclusive and prudent and categorical supervision of generative AI services, and clarifies the overall requirements for the provision and use of generative AI services. Specific measures to promote the development of generative artificial intelligence technology are proposed, and the requirements for training data processing activities and data labeling are clarified.
The overall principle tone of the Measures is also mainly encouraged, while emphasizing inclusiveness and prudence, and firmly optimistic about multiple investment opportunities brought by the development of generative artificial intelligence. At present, there are many 100 billion parameter-level large models in China, and it is expected that with the increase of model parameters and data, the cost of single model training will continue to increase, and the demand for computing power chips required for corresponding training will also increase. At the same time, with the implementation of the Measures, it is expected that the subsequent large-scale model commercialization scenarios will gradually land.