China’s AI research still behind the west, but not for long
Government investment has led to a 150% increase in Chinese AI papers since 2007
Significant government investment has allowed for a surge in the number of AI research papers being published in China, helping the country to close the gap between western rivals in terms of research contribution.
China is still lagging behind the US and UK in terms of significant contributions and overall development of AI in the country, according to the latest AI Index report.
However, China is catching up quickly. In the study, which covered a vast range of data points, it was found that China edged out the US in the number of published research papers for AI in 2018.
Europe still took dominance with 28% of all authored papers, however, China now sits at 25%, following a 150% increase of published papers between 2007 and 2017, led largely by its government.
In 2017, the Chinese government produced nearly 4x more AI papers than private Chinese corporations. China has also experienced a 400% increase in government-affiliated AI papers since 2007, while corporate AI papers only increased by 73% in the same period.
However, this surge in the number of papers produced has drastically reduced quality, with China scoring lower than Europe and the US based on the influence of published research.
Calculated using the "field-weighted citation impact (FWCI)", it was found that the US is the most influential region in the field scoring just under 2, with Europe scoring slightly under 1.5 and China nearly hitting 1.
China ahead in language processing
Another of the analysed data points was the specific capabilities in which the regions excel in the field of AI. This metric showed the number of businesses that have embedded AI capabilities in at least one business unit.
China lagged behind its western counterparts in machine learning but excelled in a variety of areas including autonomous vehicles, natural language (NL) generation and NL text understanding as well as conversational interfaces and computer vision.
Europe dominated robotic process automation while North America took a slight lead in machine learning, although all three regions scored similarly for this as well as physical robotics.
However, it wasn't all good news for the west, which continues to struggle with its diversity figures. In a data set of select universities in Europe and North America, it was shown that on average 80% of AI professors in leading universities including UCL, Oxford and Stanford are male.
In an effort to control the diversity to more equal levels, action groups such as Women in Machine Learning (WiML) and AI4ALL have seen incredible success with their programs aiming to encourage women and young people of less represented backgrounds to get into AI. WiML saw a 600% increase in workshop attendance and AI4ALL saw an increase of 900% for its educational program compared to 2015.
One of the reasons for China's massive surge in AI performance in recent years is due to the government backing provided to it.
In previous years, the US relied heavily on significant investment from private firms. Companies such as Amazon invested $16.1bn into the industry and Alphabet also injected $13.6bn, while public funding was comparatively low. For example, the National Science Foundation only allocated $5.3bn in its 2019 budget.
According to CB Insights, China accounted for 48% of the world's total AI startup funding in 2017, compared to 38% for the US. Billions more in funding has come from venture capitalists in recent months too, propelling China to the forefront of the AI race.
Top 5 challenges of migrating applications to the cloud
Explore how VMware Cloud on AWS helps to address common cloud migration challengesDownload now
3 reasons why now is the time to rethink your network
Changing requirements call for new solutionsDownload now
All-flash buyer’s guide
Tips for evaluating Solid-State ArraysDownload now
Enabling enterprise machine and deep learning with intelligent storage
The power of AI can only be realised through efficient and performant delivery of dataDownload now