Magic AIGC | The summit meeting is over, then what?

82782f22143a0974e1c02272a50b04ab.gif

551fdcab36e23764b8249facd2a7ccd8.jpeg

Image source: Generated by Unbounded AI

From July 6 to 8, the 2023 World Artificial Intelligence Conference was held in Shanghai. This industry event sponsored by the National Development and Reform Commission, the Chinese Academy of Sciences, the Ministry of Industry and Information Technology and other units has completely pushed the popularity of AI and large language models to historical highs.

133 themed forums with 177,000 participants, unrivaled.

At the meeting, more than 30 well-known domestic enterprises "attacked collectively" and rushed to show their large-scale model layout and achievements. Pinwan published an article "The large models of the World Artificial Intelligence Conference are here, let you see enough at one time":

There is no word in the full text, 60 pictures, all of which are the three words "large model" taken on the spot , The reading volume exploded.

147409f51aafd5f0dfdd357dcc5f0549.png

Some participants ridiculed after the meeting: I think it can be renamed "2023 World Large-scale Model Exhibition". Within a week, I didn't want to see the word "big model" anymore, my head was getting bigger.

01

The group of "models" danced wildly, and
the names of mythical characters are not enough?

How many large models will appear in this conference? According to incomplete statistics from the media, there are more than 30.

Including Huawei's Pangu, Baidu's Wenxinyiyan, Tencent's Hunyuan, HKUST Xunfei's Xinghuo, Alibaba Cloud's Tongyi Qianwen, JD Cloud's Yanxi, Netease's Danqing, and 360's Brain , the mountains and seas of Yunzhisheng, Yuyi of China Telecom, Mencius of Lanzhou Technology, the sequence monkey who went out to ask questions...

There are many names, such as Bole, Shangjia, Yandi, Qiusuo, Wuya...

This conference is still going on. Many new products have been released, such as the "Scholar" of the Shanghai Artificial Intelligence Laboratory, which claims to have completely defeated ChatGPT in many tests. SenseTime is new every day, and this is the Nth model it released. Youkede and Daguan Data launched Cao Zhi, which is said to be the first vertical industry-specific, independent and controllable GPT large language model in China.

With so many so-called large models, it’s no wonder that some guests at the conference bluntly said: The name is so cursive, it’s really a "thousand-model war", and I am afraid that the names of mythical figures will not be enough in the future.

This time, a big player who has never been absent from the hot spot did not show up, and that is Ant Group. It is said that its chastity model is still under development, but it has come up with the "Ant Jian AI Security Detection Platform", which can detect AI vulnerabilities and risks.

On the one hand, there is a flood of large models. On the other hand, these so-called general large models, or large models in a wide variety of vertical domains, have very single applications that can be displayed and experienced. They are mainly dialogue questions and answers. , sound synthesis, language translation and other applications. The repetition rate is called a high.

No wonder Huawei made a "big news" during the conference, and the media frantically forwarded it, and practitioners said it with emotion.

Zhang Pingan, executive director of Huawei and CEO of HUAWEI CLOUD, said when introducing his large model: "The Pangu large model does not compose poetry, nor does it have time to write poetry, because it needs to go deep into all walks of life, and let AI give value to all walks of life.

" During the conference, I also witnessed an embarrassing scene. The host who randomly interviewed at the scene was very curious about a certain AI painting model, hoping that it would paint on the spot based on a poem, but the staff stuck it to her ear and said embarrassingly: Customized , and cannot be entered arbitrarily.

02

From the block chain to the big model,
history is never-ending repetition?

Looking at today's large-scale melee, I can't help but think of the industrial blockchain wave that began in 2019-2020.

In those two years, on the one hand, various regions successively introduced guiding policies to continuously strengthen support for the blockchain. According to the statistics of the CCID Blockchain Research Institute, by the end of 2022, there have been more than a thousand blockchain-related policies issued by ministries and local governments.

On the other hand, Internet giants have launched their own blockchain underlying platforms, including Ant Group's Ant Chain, Baidu Super Chain, Tencent Zhixin Chain, JD Zhizhen Chain, Huawei Huawei Chain... This situation is exactly the same. Back then, the

district As the underlying technology, blockchain is also considered to be applicable in various industries such as finance, government affairs, medical care, and agriculture.

How is the development of industrial blockchain today? At the World Artificial Intelligence Conference, there is a sub-forum called the "Industrial Blockchain Ecosystem Forum". At the meeting, Liu Quan, deputy chief engineer of the CCID Research Institute of the Ministry of Industry and Information Technology, said that the scale, growth and important position of the blockchain industry seriously do not match, and the current industry scale is less than 10 billion yuan.

The founder and CEO of Lingshu Technology, Lin Le, shared that without realistic application scenarios, technology is just a castle in the air. It is a difficult and correct choice for Chinese blockchain companies to penetrate into the physical business field. The "industrial" blockchain is only way.

This shows from the side that four years after the rise of the blockchain wave, landing is still a headache for blockchain practitioners.

Not objectively speaking, as a media, I do feel that the blockchain industry is very cold and cold. Will AI and large models repeat the "development track" of the blockchain: it becomes very popular, and then falls into a situation where it is difficult to implement the technology and practitioners do not make money. Will the Iphone moment of the large model advertised today be the first year of the blockchain in 2019?

This may be one-sided, but it is indeed the intuitive feeling of many participants.

03

We stand at both the apex of hype
and the beginning of an era of new technology

Whether it's ChatGPT or MidJourney, these world's top AIGC applications, from new product launches to application upgrades, domestic media can broadcast live + analyze in the first time, and explain all aspects. As long as you want, you can completely synchronize with the trend of global technology.

This kind of zero information gap makes many participants of the Shanghai Artificial Intelligence Conference think that the conference lacks bright spots, and the "body feeling" of many large models and applications is lower than expected, a bit plastic.

Netizen "Aggie" immediately posted his experience of participating in the Shanghai conference. He said: "About this World Artificial Intelligence Conference, the experience this morning can be described as the 'AI landing concept meeting'. That is to say, it is all Demo, landing No."

"There are many large models that claim to be world-leading in technology and surpass ChatGPT, but you have nowhere to experience and use them. There are also some large models that claim to be the base of industry vertical large models, but when you look at these application scenarios, It’s quite a PPT, very conceptual.”

A practitioner posted on Weibo, saying that many fields are incomprehensible and inconvenient to evaluate, but as a practitioner in the medical field, he very clearly mentioned that the entry barriers in the medical field are high , Various software applications require approval from the Food and Drug Administration. Moreover, is patient data easy to get? It is very difficult. It is difficult to face ethical challenges and comply with standards and norms.

"Without data, how do you train a large model?" He said, "I haven't felt the artificial intelligence in the medical field. From the things that have come to me so far - the disinfection robot in the operating room, the artificial intelligence high-value consumable cabinet, and the Internet The medical record filling system of artificial intelligence pre-inquiry in outpatient clinics—from my personal experience, these are still at the stage of artificial mental retardation.”

Some netizens also commented very directly: You think the big model is trying to compete for technology, but technology is not important... …

While writing this article, a colleague suddenly said in the group: "AIGC is a big show, and those who sell courses have already made a lot of money. The big model PPTs of the giants have also been released, industrial policies have been launched, and the highest-level meetings have also been released. It’s over, the executives seem to have either divorced or cashed out at the high point, and then?”

Another colleague answered: “Is it the end?”

The Shanghai Artificial Intelligence Conference closed on the afternoon of the 8th. A set of data was mentioned in many media announcements: the conference has connected 210 upstream and downstream enterprises, reached an intentional purchase amount of 11 billion yuan, promoted the signing of 32 major industrial projects, and the total investment of the project 28.8 billion yuan.

This is the real value of this conference beyond the hot AI industry, as a guest made an impressive speech: In 2023, the general-purpose large model will detonate the AI ​​storm, and the 24-year vertical model will definitely reshape thousands of industries.

Halfway through 2023, we may be at the apex of AI hype and at the starting point of a new round of technological revolution.

End this article with a very popular sentence from Tang Daosheng, Senior Executive Vice President of Tencent and CEO of Cloud and Smart Industry Business Group: General large models can solve 70%-80% of the problems in 100 scenarios, but not necessarily It can 100% meet the needs of a certain scene of the enterprise. If an enterprise conducts fine-tuning based on the large industry model and its own data, it can construct a dedicated model and create highly available intelligent services.

Looking forward to the arrival of 2024, looking forward to more applications landing.

Babbitt Park is open for cooperation!

58949ed1486705649ed7026faf56d6f4.png

6a797942d92bc66207da4228bc5a0c86.jpeg

dafc7187a6b3b2c0e3c2ff50a37c166b.gif

Chinese Twitter: https://twitter.com/8BTC_OFFICIAL

English Twitter: https://twitter.com/btcinchina

Discord community: https://discord.gg/defidao

Telegram channel: https://t.me/Mute_8btc

Telegram community: https://t.me/news_8btc

89126188c55d14d31f2f46517a93ab00.jpeg

Guess you like

Origin blog.csdn.net/weixin_44383880/article/details/131671704