On May 22, it is reported that the US group has formed a team to develop a large model, which has been secretly developed for more than two months.。

Internally set up a large model research and development team of Meituan, foreign investment Wang Huiwen, Wang Xing vowed to be an AI trendsetter？

According to media reports, S-team, the highest decision-making body headed by Meituan founder Wang Xing, attaches great importance to the development of large models, and every week or two asks the head of the algorithm team about the progress of large models.。It is reported that in the context of the Internet "drawdown," the team reverse expansion, algorithms, products and other positions are in the recruitment.。

It is reported that for the recent formation of a large model team, Meituan believes that "the big model is still in the window period, but also the best time point for Internet companies to accelerate their development, which is crucial to Meituan's strategy and growth."。"

In addition to working on research and development, Meituan is also seeking to set up a separate department to assist in the commercialization of large models.。It is reported that Meituan's large model business may appear in the form of Meituan Assistant in Meituan App, or it may come out in the form of a separate App.。

Judging from the current information, the Meituan model is still in the exploratory stage, and the direction of subsequent commercialization is still uncertain, which may be the reason why Meituan is "secretive."。

As early as early as March, Wang Xing said on his social account: "The AI model makes me both excited about the huge productivity that will be created, and worried about its future impact on the whole world."。Lao Wang and I have been together on the road to entrepreneurship for nearly two decades, and since he is determined to embrace this big wave, I must support。Personally, I will participate in the Series A investment of Lao Wang Ventures' Light Years Away 'and serve as a director.。"The old king in his mouth refers to Wang Huiwen.。

As the co-founder of Meituan, Wang Huiwen has always been an entrepreneur.。He once founded Xiaonei Network with Wang Xing, later Renren Network, and later Taofang Network, and joined the US Mission in 2010, once again fighting side by side with Wang Xing.。He was in charge of marketing, bike sharing and other businesses in Meituan, and in 2020 reduced his holdings of 2 million Meituan shares and announced his retirement.。

But this AI wave has aroused the interest and fighting spirit of this entrepreneurial talent.。On February 13, Wang Huiwen announced that he would form a team and build a Chinese version of OpenAI, saying that he had set up Beijing Lightyear Beyond Technology Co., Ltd. (hereinafter referred to as "Lightyear Beyond"), which invested US $50 million, accounting for 25% of the shares, and that the remaining 75% of the shares would be used to invite top R & D talents.。He also claimed that the next round of financing has been subscribed by the top VC 2..300 million dollars。According to the company, the company was founded in 2018 and is 100% owned by Wang Huiwen.。The company also took a stake in Beijing First-Class Technology Co., Ltd. (hereinafter referred to as "First-Class Technology") last month, with a shareholding ratio of about 47%, First-Class Technology was established in 2017, is an AI architecture startup, with the core product OneFlow deep learning framework.。But for investments beyond light years, Wang Xing does so in his own name。

While building a large model team in the US regiment, while supporting old comrade-in-arms Wang Huiwen, the "two-handed grasp" highlights Wang Xing's emphasis on artificial intelligence.。

 

Big guys are bullish on AI, and the big model track is already running.

According to media statistics, at least 40 companies and institutions in China have released large model products or announced large model plans so far.。

Last week, the 7th World Intelligence Conference was held in Tianjin, 360 Group founder Zhou Hongyi delivered a speech at the conference, and showed two large model products "360 brain" and AI graphics tool "360 Hongtu."。Zhou Hongyi said in his speech that China will not have only one big model in the future, each city and each government department will have its own proprietary big model.。The peak of digitalization in the future must be intelligence driven by large models, and the exponential leap in artificial intelligence will bring huge development opportunities.。He believes that big models may be a standard technology in the future, just like knowledge management.。There are more enterprises to participate in the competition and investment, in order to achieve a breakthrough。But at the same time, he stressed that at this stage, China's large model products and ChatGPT still have a big gap, should be down-to-earth to make up for the shortcomings, he believes that the next two to three years, China's large model development has a great opportunity.。

Although Zhou Hongyi is very optimistic about AI, 360 is a bit stretched。The financial results disclosed last month showed that 360 recorded a loss of 2.2 billion yuan in net profit last year.。For AI that needs a lot of "burning money," the pressure on 360 funds is not small.。

Internet giant Tencent, which is not short of money, is also making big models.。It is reported that Tencent has said that its AI big model hybrid has made corresponding progress in data, model sequence, infrastructure and other aspects.。"The big model is a growth multiplier for the industry and Tencent. It can improve service quality, increase liquidity efficiency, and reduce costs。"

Ma Huateng, founder of Tencent, recently said in response to a question about AI: "At first we thought this was an opportunity that the Internet has not seen in a decade, but the more we think about it, the more we think it is an opportunity that has not been seen in hundreds of years, similar to the industrial revolution that invented electricity, so we think AI is very important.。But this does require a lot of accumulation.。Internet companies have a lot of accumulation, are doing, we are also immersed in research and development, but do not rush to finish early, the semi-finished products out to show。For the Industrial Revolution, getting the light bulb out a month earlier is less important over a long time horizon.。The key is still to get the underlying algorithms, computing power and data down to earth, and more importantly, the scene landing。"

 

Baidu, which lays out AI early, is also a force to be reckoned with.。Previously, Baidu withstood the pressure to take the lead in releasing "Wen Xin Yi Yan," becoming the first technology company in China to release products similar to ChatGPT.。In research and development, Baidu also insists on "burning money."。Baidu's R & D expenses in 2022 will be 233.1.5 billion yuan, accounting for about 19% of total revenue。

At the recent sixth anniversary of Baidu's new mission and Baidu Pride Awards, Baidu founder Robin Li said: "Two years ago, we realized that artificial intelligence technology has undergone a directional change, from discriminative artificial intelligence to generative artificial intelligence, AIGC has since become a direction of our focus, resources are gradually tilted in this direction.。"

He also said, "In the real AI era, new applications will be developed based on large models, just like a large number of applications based on Android and iOS emerged more than a decade ago."。The big model is Game Changer, which will completely change the rules of the game in cloud computing。In the past, cloud computing mainly sold computing power, speed, storage, today, customers buy cloud services, is to see the framework is good, the model is good, rather than just how the computing power。"

In addition to the major domestic private technology companies, the "national team" also joined the track.。

HKUST Xunfei released the "Xunfei Spark Cognition Model" on May 6.。According to the company, the Xunfei Spark cognitive model has seven dimensions of capabilities, including text generation, language understanding, knowledge answering, logical reasoning, mathematical capabilities, code capabilities, multi-modal capabilities, and has been applied in education, office, automotive, digital employees and other industries.。At the press conference, Liu Qingfeng, chairman of iFLYTEK, announced that the first batch of more than 3,000 enterprise developers from 36 industries will be connected to the Spark model.。

iFLYTEK is ambitious about the model。It is reported that the follow-up Xunfei Spark cognitive model will achieve three stages of upgrading and improvement.。In the first phase, it is expected to break through the open question-and-answer process in June, significantly improve the ability of multiple rounds of dialogue, and upgrade the mathematical ability.。In the second phase, the breakthrough code capability and multimodal interaction capability will be officially opened to customers in August.。In the third stage, the general model will be benchmarked to ChatGPT in October, with Chinese surpassing the current version of ChatGPT, English at a comparable level, and industry-leading in education, healthcare and other fields.。

Not only iFLYTEK, there are now a number of domestic enterprises ambitious, aiming at the "Chinese version of ChatGPT" throne。The big model track is now "running," the crowd is competing, who can become a domestic AI trendsetter, time will tell us the answer.。

 

Eagle Statement: This article is for reference only and does not constitute personal investment and operational advice.。Special reminder, the article is original content, without permission may not be reproduced。