i3 - May/June 2018 - 23

REASONS FOR CHINA'S AI STRENGTHS
a financial analysis firm, found that in 2017, Chinese
AI startups took in nearly half (48 percent) of funds
going to such firms worldwide. That was the first
time that U.S. developers were not number one in
share of AI investment dollars.
Another metric is the number of patents issued,
which reflect a country's research and development
pusuits. Although the patent filing and authorization process have many hurdles regarding schedules, the CB Insights assessment shows that China
patents are "surging far ahead of patents" in the U.S.
in categories such as artificial intelligence, machine
learning and deep learning. For example, China had
about six times as many patents published on deep
learning than did the U.S. (652 versus 101) from
2013 to 2017. CB Insights points out that China's
focus on facial recognition and AI chips fit into the
government's country-wide surveillance plans.
At the annual conference early this year of the
Association for the Advancement of Artificial
Intelligence, presenters from the U.S. and China
dominated the agenda. But while the number of
research papers from U.S. participants (34 percent
of total papers) had dropped by about seven percent
during the previous five years, the number of Chinese
papers (23 percent of papers) jumped 13 percent.

MAGIC MIRROR TO VIDEO STREAMING

While smart cars and remote health care monitoring/treatment are at the core of many AI developments, backers see opportunities in other major
business categories.
For example in the retailing sector, a "smart fitting room" is helping apparel customers make
clothing selections based on a variety of factors.
Using facial recognition technology, the AI system
helps customers assess options as they try on clothes
in front of a "magic mirror." The device, being tested
by Besteller A/S, the company behind the fashion
brands, calculates age, gender, climate and other
factors to present alternative clothing choices which
appear in the mirror, without requiring customers
to actually change outfits.
Alibaba Group Holding Ltd. offers a similar electronic mirror that allows customers to chose Shiseido
lipsticks and complete a purchase by tapping the screen.
iQiyi Inc., a Baidu subsidiary which operates
China's largest video streaming service, is offering
an AI system to help viewers pick shows they want
to see. Executives at the company even used iQiyi
Brain to identify programs to include in its roster,
opening the prospect for corporate media leaders to
turn to AI for creative decisions.

Many factors are cited for China's ability
to develop AI systems quickly:

Massive data collection capabilities:
China is leveraging its status of having the world's
largest population. Since AI and machine learning systems are often built on processing vast
quantities of data, the Chinese market generates
unmatched numbers used to create AI systems.
"A large amount of data makes a large amount
of difference," Sinovation Ventures CEO Kai-Fu
Lee said recently.

Widespread use of mobile:
Chinese customers' extensive use of wireless
devices means that data is coming from many
locations, again adding to the building blocks on
AI learning modules. Mobile payments in China are
about 50 times greater than in the U.S., according
to the Financial Times - suggesting a propensity to
adopt tools that are fundamental in AI systems.

Few privacy restraints on data collection:
Unlike American and European developers, Chinese
AI ventures can collect considerable personal
data from early users - reflecting both the political
structure of the country and individuals' willingness
to share such personal information.

Dealmaking on a global scale:
Baidu's Apollo autonomous driving platform represents an example of the big AI company's effort to
work worldwide. At CES, Baidu revealed relationships
with four primary computing platforms including
NVIDIA, Intel (both from Silicon Valley), NXP (the
Dutch semiconductor manufacturer) and Renesas
(Japan's large automotive semiconductor maker).

MAY/JUNE 2018

i3 - May/June 2018

Table of Contents for the Digital Edition of i3 - May/June 2018