Stanford 2023 AI Index Report is out! China dominates the top AI conferences, with the Chinese Academy of Sciences ranking first in the world in terms of published papers

Today, Stanford released the 2023 AI Index Report.

It is worth noting that the Stanford AI Index report lists the top ten institutions in the world in terms of "AI paper publication volume", and 9 of them are all from China. , have caught up with MIT.

They are: Chinese Academy of Sciences, Tsinghua University, University of Chinese Academy of Sciences, Shanghai Jiao Tong University, Zhejiang University, Harbin Institute of Technology, Beihang University, University of Electronic Science and Technology of China, Peking University, and MIT.

This year’s report is mainly divided into eight major sections: research and development, technology performance, artificial intelligence technology ethics, economy, education, policy and governance, diversity, and public opinion.

The following content extracts several key points of the report.

The two countries rank first in paper cooperation in the world

From 2010 to 2021, although the pace of cross-border cooperation in AI papers has slowed down, since 2010, the United States The number of artificial intelligence research cooperation with China has increased by approximately 4 times, which is 2.5 times greater than the total number of cooperation between China and the UK.

However, from 2020 to 2021, the total number of cooperation between the two countries only increased by 2.1%, which was the smallest year-on-year growth rate since 2010.

Additionally, the total number of AI papers has more than doubled since 2010. It has increased from 200,000 articles in 2010 to nearly 500,000 articles in 2021 (49,601).

In terms of the types of AI papers published, in 2021, 60% of all published AI papers are journal articles, and 17% are Conference papers, 13% come from repositories.

While journal and repository papers have grown 3-fold and 26.6-fold, respectively, over the past 12 years, the number of conference papers has declined since 2019.

Pattern recognition, machine learning and computer vision are still hot topics in the field of artificial intelligence.

China still leads the way in terms of the total number of papers in journals, conferences and repositories.

The United States still leads in AI conference and repository citations, but these leads are slowly eroding. Despite this, the majority of the world's large language models and multimodal models (54% in 2022) are produced by US institutions.

China dominates the top AI rankings, but the number of citations is lower than that of the United States

China has always maintained its lead in publishing AI journal papers status, 39.8% in 2021, followed by the European Union and the United Kingdom (15.1%), and then the United States (10.0%).

Since 2010, the proportion of citations of Chinese artificial intelligence journal papers has gradually increased, while the EU, UK, and United States have all declined. China, the European Union, the United Kingdom, and the United States account for 65.7% of the total global citations.

So, what is the situation of papers published at the world’s top conferences?

In 2021, China has the largest share of papers published at the top global AI conferences with 26.15%, followed by the European Union and the United Kingdom with 20.29%, and the United States with 17.23 % ranked third.

Judging from the number of citations of top conference papers, although China is highly productive, its number of citations is lower compared to the United States. The number of citations of top conference papers in the United States was 23.9%, and that in China was 22.02%.

It can be seen from the side that China publishes the largest number of papers, but the quality is not as high as that of the United States.

The United States leads the world in submissions to AI paper repositories, with 23.48%. China is the lowest, 11.87%.

9 institutions in China, AI paper publishing catches up with MIT

In 2021, the total number of published papers is among the top ten institutions in the world , China accounts for 9. The total number of papers published by different institutions is as shown below. MIT ranks tenth, publishing 1745 papers.

In terms of the field of computer vision (CV), China’s ten institutions rank among the top ten in the world. They are: Chinese Academy of Sciences, Shanghai Jiaotong University, University of Chinese Academy of Sciences, Tsinghua University, Zhejiang University, Beihang University, Wuhan University, Beijing Institute of Technology, Harbin Institute of Technology, and Tianjin University.

In the field of natural language processing (NLP), it is different.

The top ten institutions/companies in the world are: Chinese Academy of Sciences, Carnegie Mellon University, Microsoft, Tsinghua University, Carnegie Mellon University-Australia, Google, Peking University , University of Chinese Academy of Sciences, Alibaba, Amazon.

The rankings in the field of speech recognition are as follows:

Industry leads academia

Among the important artificial intelligence machine learning systems released in 2022, language systems account for the most, with 23, which is 6 times the number of multi-modal systems.

In terms of paper output, industry is ahead of academia.

Until 2014, most important models were published by academia. Since then, industry has turned around. By 2022, 32 important machine learning models will be born in industry, while only 3 will be in academia.

It can be seen that compared with non-profit organizations and academia, building state-of-the-art artificial intelligence systems increasingly requires large amounts of data, Computer capabilities and financial resources, and industry participants certainly have more financial resources to do this.

In 2022, the United States produced the largest number of important machine learning systems, with 16, followed by the United Kingdom (8) and China (3).

Furthermore, the United States has surpassed the United Kingdom and the European Union and China in terms of the total number of significant machine learning systems created since 2002

Looking at the distribution of researchers behind these important AI systems by country, the United States has the largest number of researchers, 285, which is more than twice that of the United Kingdom and nearly six times that of China.

LLMs are getting bigger and bigger, and the computing power is more expensive

Large language and multi-modal models, sometimes called base models , is currently an emerging and increasingly popular type of AI model that is trained on large amounts of data and suitable for various downstream applications.

Large-scale languages ​​and multi-modal models such as ChatGPT, DALL-E 2, and MakeA-Video have demonstrated impressive capabilities and are starting to be widely deployed in the real world.

The country affiliations of the authors of these models were analyzed and the majority of these researchers were from US institutions (54.2%).

The Stanford AI Index report also lays out a timeline for the release of large language and multi-modal models.

#Large language models are getting bigger and more expensive.

The first large-scale language model, GPT-2, was released in 2019, with 1.5 billion parameters and a training cost of about US$50,000. Google PaLM is one of the large language models launched in 2022, with 540 billion parameters and a cost of up to $8 million.

In terms of parameters and training costs, Palm is 360 times larger and 160 times more expensive than GPT-2.

It’s not just PalM, but large language and multimodal models as a whole are getting bigger and more expensive.

For example, Chinchilla, a large-scale language model launched by DeepMind in May 2022, is estimated to cost US$2.1 million, while training of BLOOM costs approximately US$2.3 million.

Over time, the progress of GAN in face generation, the last image was generated by Diffusion-GAN, this model in STL The latest SOTA was obtained on -10.

Last year, with OpenAI’s DALL-E 2, Stability AI’s Stable Diffusion, Midjourney, Meta’s Make-AScene, and Google’s Imagen With the release of the model, the text-to-image generation model has gradually entered the public eye.

As follows, enter the same prompt, "A panda plays the piano on a warm Parisian night", respectively, from three publicly accessible programs: DALL-E 2, Stable Diffusion and Midjourney. Image generated by AI text-to-image system.

Among all recently released text-to-image generation models, Google’s Imagen performs best on the COCO benchmark.

This year, the Google researchers who created Imagen also released DrawBench, a more difficult text-to-image benchmark designed to challenge increasingly powerful text-to-image models.

In addition, the report also introduced that there are some biases in the current generative AI model. For example, when prompting the CEO to DELLE-2, everyone seemed to take crossed arms in a confident pose.

In Midjourney, when prompted to generate "influencers" it generates 4 images of older looking white men .

For the complete report content, please see:

​https ://aiindex.stanford.edu/wp-content/uploads/2023/04/HAI_AI-Index_Report_2023.pdf​

The above is the detailed content of Stanford 2023 AI Index Report is out! China dominates the top AI conferences, with the Chinese Academy of Sciences ranking first in the world in terms of published papers. For more information, please follow other related articles on the PHP Chinese website!