The 'Battle of 100 Models' has begun, and AI large models have completely set off a new wave of applications?

The development of technology has now become an indispensable part of our lives. From smartphones to artificial intelligence, technology is changing the way we live our lives at an alarming rate. Whether at work or in daily life, technology has brought us many conveniences and innovations. The popularity of smartphones allows us to stay in touch with others, obtain information, and even conduct various online transactions anytime and anywhere. We no longer need to rely on traditional communication methods, but can easily complete various tasks through mobile phones. In addition, smartphones also provide various applications that can help us manage time, health, finance and other aspects. The development of artificial intelligence has brought us unprecedented opportunities and challenges. Through machine learning and deep learning technologies, artificial intelligence can simulate human thinking and decision-making processes. This allows us to develop innovative products and services such as smart assistants, self-driving cars, and smart homes. However, the development of artificial intelligence has also raised a series of ethical and privacy issues that require in-depth thinking and discussion. In short, the development of science and technology has brought us huge opportunities and challenges. We should actively embrace technology, but at the same time be wary of the negative impacts that technology may bring. Only with the harmonious coexistence of technology and human beings can we achieve a better future

The era of AI large models is accelerating.

For many technology giants and unicorn companies, if they do not get involved in the field of artificial intelligence, they seem to be left behind by the times. The best proof is the launch of large artificial intelligence models. The current industry development trend shows that more than 130 large artificial intelligence models have appeared in China in a short period of time, and vertical segmentation has also emerged based on different strengths, directions and priorities

This sudden and all-out "Battle of 100 Models" is coming with ferocity. It not only tests the technical foundation of technology and Internet companies, but also brings new challenges: how to stop "painting the pie" in the air and strive to apply in applications Leading the way at all levels? Under such collective thinking, AI applications are about to enter an era of explosion.

We can see that based on large AI models, the segmented tracks of generative AI applications are constantly being developed or expanded. AI office software, AI creative tools, AI enterprise services, AI network security, AI finance, medical care, education, etc. are in a rapidly changing development trend. It is foreseeable that this new wave will have far-reaching impacts!

"Battle of 100 Models": Involution is coming

Recently, important domestic news about large-scale artificial intelligence models has been coming one after another

For example, on August 31, 11 domestic large-scale models have successively passed the "Interim Measures for the Management of Generative Artificial Intelligence Service" for registration, and have successively opened their services to the public. To put it simply, many large models were mainly in internal testing before, but now after passing the registration, both enterprises and individuals can use them with confidence.

These 11 domestic large-scale models are currently the leaders in the country, including Baidu’s Wenxinyiyan, Douyin’s Skylark Large Model, Chinese Academy of Sciences’ Zidong Taichu, SenseTime’s SenseChat, Huawei’s Pangu Large Model, Tencent’s Hunyuan large model, iFlytek’s Spark large model, etc. As of now, they are basically open.

It can be said that August 31, 2023 is a key time node worth remembering in the history of the development of domestic large models. Because starting from this day, the "gear of destiny" belonging to the large domestic AI model has really started to turn!

In addition, the second batch of large-scale models for registration are also being rapidly advanced. On September 1, the Cyberspace Administration of China released the “Information Announcement on the Registration of the Second Batch of Deep Synthesis Service Algorithms.” This announcement shows that a total of 110 deep synthesis service algorithms have been registered

The opening of domestic large models is of great significance. Baidu founder Robin Li once mentioned that when Wen Xinyiyan opens its services to hundreds of millions of Internet users on a large scale, it will be able to obtain a large amount of real-world manual feedback, which will further improve the basic model and build a faster Iterate. For other large domestic models, openness will naturally have the same effect

Facing the future, with the opening up of domestic large-scale models, the "Battle of Hundreds of Models" will become more intense. What you need to know is that although there seem to be a large number of large AI models in the field of domestic large models, they have not yet truly been baptized by market rules. They are still in the stage of staking land and the pattern has not yet been fully formed. This also means that everything is possible. possible.

From a data perspective, since last year, there have been nearly 100 financing cases in the large-scale model industry in China. In terms of the investment valuation of the entire industry, it is estimated that by 2023, the new investment in this industry will be between 15 billion and 20 billion yuan. In the long run, the massive investment is well worth it. According to calculations by Large Model House, the global large model market is expected to reach US$21 billion by 2023, and reach US$109.5 billion by 2028

Faced with a promising future, in the competition of large AI models, participants will inevitably go all out to try to stand out in the fiercely competitive market and become the ultimate winner. This also shows that the phenomenon of involution is inevitable and has even arrived!

Regarding the competitive situation in the field of domestic large models, Jiang Han, a senior researcher at Pangu Think Tank, put forward his own views. He believes that there are currently many large models that represent market compliance through registration and maintain sufficient competition. The diversified characteristics of the domestic large model field are still very obvious, which is in line with the current initial stage of market development

Domestic large models, let a hundred flowers bloom. General large models, industry large models, vertical large models... Major technology and Internet companies have chosen the model types they are best at to start, and competition has become fierce. Such competition and involution are obviously beneficial to the industry and the public.

Even with the arrival of the reshuffle period in the future, it is unlikely that one large AI model will dominate the world, and there will still be a situation where all the heroes are vying for the top spot. It’s just that some large AI models will focus on market segments and pursue “small but beautiful”

The rapid development of artificial intelligence will trigger a new wave of applications

What you must know is that the large AI model is the core of generative AI. A wide variety of large AI models with different performances can stimulate the potential of generative AI and lead to a broad application ecosystem.

If the large AI model is regarded as infrastructure, then more vertical models in professional fields can be incubated on top of it to serve the digital transformation of various industries. As domestic large-scale models become fully mature and open, the public will also have access to more choices and possibilities. Enterprises and individuals can explore new application scenarios through generative AI, thereby bringing innovation and opportunities to enterprise development, daily life and work

Wu Tian, ​​vice president of Baidu Group and deputy director of the National Engineering Research Center for Deep Learning Technology and Applications, said that the phenomenon of large AI models blooming indicates a trend that generative AI may have a profound impact on industrial structure and society in the future. This is just the beginning, and AI’s penetration into the industry is deepening. There is no doubt that generative AI will drive global productivity innovation and reshape business models

The application of AI in different segments is being demonstrated in an intuitive way. Especially in the fields of office software, creative tools, enterprise services, network security, finance, medical care, education and other fields, AI is becoming an important emerging growth area

Taking AI medical care as an example, according to market research consulting firm MarketsandMarkets, the global medical large model market will reach US$3.8 billion by 2025, and by 2030, this number will exceed US$10 billion. Giants such as SenseTime, iFlytek, JD.com, Baidu, and Tencent, as well as Internet technology companies such as Wanda Information, Weimai, and Dingdang Health, are all paying close attention to AI medical care and are making big moves

For example, SenseTime Technology has released an upgraded version of the "SenseCare Smart Hospital" comprehensive solution, creating a new paradigm for large-scale medical model applications. Tencent Health announced a large medical model on September 8, launching a new multi-scenario AI product matrix such as intelligent question and answer, family doctor assistant, and digital medical imaging platform. Baidu also officially released China's first "industrial-level" medical model on September 19 - the Spiritual Medicine Model, and announced that it will be open to the upstream and downstream industries of big health.

Facing AI office, many companies have entered the game. For example, on September 5, Kingsoft Office officially announced that WPSAI is officially open to the public. AI capabilities will be the first to be applied in WPS smart documents, and more WPS AI capabilities will be opened in other components. iFlytek has even empowered its smart office notebooks and other products with large-scale Spark models, demonstrating the charm of AI in improving work efficiency.

As for AI education, it is also a position where generative AI plays an important role. For example, compared with other peers that focus on the C-end of education, Baidu has taken a different approach. Just on September 15th, Baidu Marketing released the first AIGC education industry marketing solution, including the three major marketing tools of "Qingge", "Qingduo" and "Merchant Bot" and the two "Lexiotong" and "Education Business Opportunity Treasure" Big innovative products aimed at improving quality and efficiency throughout the entire advertising and marketing link in the education industry.

Of course, generative AI has actually been successfully implemented in thousands of industries and has become a catalyst for industry innovation. Overall, generative AI applications can be divided into four categories according to application fields: tools, general software, industry software, and intelligent hardware.

Among them, the tools mainly include chat robots, search engines, text tools, Al painting and coding tools, etc. The application fields are mainly concentrated on the C side, with a high degree of homogeneity and a high degree of dependence on the capabilities of the underlying model. General software mainly covers office software, enterprise services, IT operation and maintenance, software development, network security, data intelligence, etc. Benchmark products have appeared on various tracks, most of which are in the form of intelligent assistants, which are about to enter the key to commercialization. stage.

The scope of industry software involves finance, medical care, education, industry, games, law and other industries, and the differences between these industries are very large. In the future, fully tapping the value of industry data will become the key to competition. Intelligent hardware includes cars, robots, smart terminals, etc. The current challenge lies in the perception layer and decision-making layer, which still require further breakthroughs

As domestic large-scale models increasingly show great potential, 2023 is destined to be a breakthrough year for generative AI at the application level. The possibilities that generative AI brings to many industries are full of unlimited imagination. The constantly evolving and evolving generative AI will revitalize thousands of industries and achieve vigorous development through continuous innovation.

The above is the detailed content of The 'Battle of 100 Models' has begun, and AI large models have completely set off a new wave of applications?. For more information, please follow other related articles on the PHP Chinese website!