China is close to becoming the world’s leader in artificial intelligence, according to conventional wisdom.
But Jeffrey Ding, leader of all things China at the Center for the Governance of AI at Oxford University’s Future of Humanity Institute, has a much different take: China’s prowess in artificial intelligence is exaggerated.
Ding’s take on China and A.I. is a counterpoint to the belief that China and the U.S. are embarking on an A.I. arms race that echoes the Cold War. Misinformation could prompt the U.S. to go overboard with its A.I. policy related to China or implement misguided policy, Ding says.
In an interview with Fortune, Ding explained that much of what is written about China’s multi-billion dollar push into A.I. often seems like it’s written in a “vacuum.” There’s little context or comparison between China’s A.I. abilities and those of other countries.
For instance, some reports indicate that China leads the U.S. in the number of A.I.-related patents globally and research papers. No one bothers to mention that researchers outside of China more frequently cite U.S. patents and papers in their own studies, indicating that the Chinese patents and papers may be of lesser quality, Ding says.
It’s these kinds of observations that make Ding’s weekly ChinAI Newsletter widely read among China and A.I. experts. His analysis is informed by a network of sources in China who help him translate Chinese A.I. research papers and related writings.
Another of Ding’s contrarianism targets the assumption that China’s vast population provides a huge workforce of A.I. specialists. Yet quantifying the size of that workforce is difficult because there’s no standard for doing so, he says. Some workers in China who claim A.I. expertise may only have associate or technical certificates, meaning they likely lack the skills of people with more advanced engineering degrees.
China has also been singled out as a leader because it has compiled a database of over one billion facial images. In theory, it’s a valuable tool for training facial-recognition technologies, a key subset of A.I.
Less appreciated is the fact that the FBI has its own database of over 640 million faces, according to a Government Accountability Office report. Like China, the U.S. is also well on its way to track people’s identities based on their faces, despite concerns by human-rights activists.
Assumptions about privacy expectations in China are another source of misunderstanding in the West, Ding says. In short, because of China’s authoritarian government, Westerners believe that the country’s citizens don’t value privacy.
Ding argues that this thesis overlooks a “crucial distinction” between civil liberties (i.e. privacy related to the government) and consumer privacy (how companies handle personal information). Chinese consumers, he said, are wary of companies that are lax in safeguarding their personal data and are concerned about their data being stolen.
“If you want sustainable A.I. development, you have to get better at protecting data,” Ding said.
Recently, Ding testified during a U.S. congressional hearing about China’s pursuit of A.I. and its implications on U.S. national security and technology policies. His conclusion: Despite the belief by some experts that China may overtake the U.S. in A.I., the reality is much more nuanced.
Jonathan Vanian
@JonathanVanian
jonathan.vanian@fortune.com
EYE ON A.I. NEWS
Where are my voice records? Amazon said it keeps the transcripts and voice recordings of people’s interactions with the Alexa voice-activated digital assistant indefinitely and will only delete the information if users manually delete it themselves, CNET reported. But the report noted that “there are still records from some conversations with Alexa that Amazon won't delete, even if people remove the audio.”
Is this what Facebook thinks about my photos? A recent Facebook outage publicly revealed how the company’s A.I. systems automatically tag user photos with words to describe the photos, the Verge reported. Because of the outage, some Facebook users may have seen words like “people smiling, people dancing” or other “tags” in place of their personal photos, the report said.
Baidu and Intel partner on chip smarts. Intel said it's collaborating with Chinese search company Baidu to develop a specialized computer chip to speed the process of training deep-learning systems. Earlier this year, Intel said it was working with Facebook to develop a computer chip to aid with A.I. inference, which happens when an A.I. system acts on the data it ingests.
A.I. pizza party. CNN reported about research from MIT and the Qatar Computing Research Institute that described an “AI system that can look at a photo of pizza and deduce what ingredients should go on which layer of the pie.” The challenge was for the researchers to create an A.I. that could understand how to properly layer a pizza instead of merely assembling the pizza haphazardly.
HEALTHCARE A.I.
Bruce Feinberg, the vice president and chief medical officer at Cardinal Health Specialty Solutions, writes in healthcare news service STAT about A.I.’s potential impact in healthcare. Feinberg and his firm surveyed over 180 oncologists and found that most believed A.I. that automates administrating tasks would be the most helpful to their profession. “This response aligns with research we conducted last year showing that oncologists need extra hours to complete work in the electronic medical record on a weekly basis and the EMR is one of the top factors contributing to stress at work,” wrote Feinberg.
EYE ON A.I. HIRES
Avast has hired Michal Pěchouček to be the cybersecurity firm’s chief technology officer. Pěchouček is a professor at the Czech Technical University in Prague, where he leads the university’s computer science department and artificial intelligence center.
EYE ON A.I. RESEARCH
A.I. to detect the age of a brain. Researchers from Columbia University and Carnegie Mellon University published a paper about using deep learning to estimate the age of brains based on a dataset of MRI brain scans. The researchers write that in order for their techniques to be used more widely, “they must be trained on adequately diverse datasets that reflect the diversity of the populations on which the model might potentially be deployed.”
A.I.-powered stock portfolios. Researchers from Columbia University, Beijing Institute of Technology, and the University of Science and Technology of China, Hefei published a paper about using reinforcement learning techniques, in which computers learn by trial, to create an investment portfolio. The researchers claim that their techniques proved “better than the Dow Jones Industrial Average and the traditional portfolio allocation strategies.”
FORTUNE ON A.I.
Facebook and Google’s Data-Control Tactics Aren’t Just a Privacy Issue Anymore – By David Meyer
iOS 13 Will Use AR Magic to Fix the iPhone’s Most Annoying FaceTime Flaw – By Xavier Harding
White House Social Media Summit: Critics of Facebook, Google, Twitter Invited to Sound Off – By Alyssa Newcomb
BRAIN FOOD
Jumping to automated conclusions. The Economist interviewed one of the authors behind a highly-cited, but widely-misunderstood academic paper about automation and jobs. Many people misinterpret the paper as saying that advanced automation technologies will make nearly half of U.S. jobs redundant by the mid-2030s. It turns out that the authors of the paper “make no attempt to estimate how many jobs will actually be automated,” The Economist noted. Instead, the authors believe that any job loss due to automation “will depend on many other things, such as cost, regulatory concerns, political pressure and social resistance,” according to the article.
