Thirsty Giant

Soft-drink bottlers worldwide depend on diminishing supplies of clean water. Here’s how Coke copes.

Coca-Cola is the world’s most popular soft drink brand. So it’s not surprising that statistics flowing from its global-leader status can be mind-boggling. For example, in each second of every day, people consume 10,450 Coke drinks, from its eponymous sodas to waters and fruit juices. And worldwide, 1.8 billion bottles of Coke drinks are sold daily. To produce all those drinks, Coke bottlers use more than 79 billion gallons of water a year.

That last stat—roughly the amount of water that pours over Niagara Falls in a 30-hour period—is as significant as it is impressive. Water is key to the success of Coca-Cola and all the other beverage makers worldwide. It not only is the main ingredient in each bottle or can of drink produced but is also essential to their manufacturing processes, particularly in the cleaning and rinsing of equipment. “If we don’t have water, we don’t have a business. That’s pretty obvious,” says Paul Bowen, the Coca-Cola Co.’s director of sustainable operations.

Which helps explain why Coke has become a global leader in using a variety of proven, highly engineered technologies and efficient production techniques to drastically reduce its consumption of freshwater in a world where that commodity is scarce and getting scarcer. Advanced technology also allows Coke to be bottled and sold in parts of the world with substandard water purification systems.

The United Nations says an estimated 783 million people worldwide lack access to clean water. River basins are slowly drying up, and climate change, population growth, and pollution are depleting Earth’s groundwater sources, according to the World Bank. Looming water shortages are a big and growing risk for the global beverage and food industry. Coke, for instance, lists water as a major ingredient under stress in its annual financial report to the Securities and Exchange Commission.

While the industry’s products are undeniably popular, none is essential to human existence. But water is. So water scarcity also presents the industry with a tricky public-relations challenge. Perhaps it’s not surprising then that while Coke and other beverage producers continue to draw criticism for sugary drinks, their record on water use compares favorably with those of other industries. A 2015 report on the water-use records of 37 global manufacturing companies by Ceres—a Boston nonprofit that works with corporations and investors on sustainability—found that the beverage and snack-food companies were doing the best job of reducing their water footprint. They’re companies that are highly susceptible to consumer blowback for poor environmental policies, Ceres notes.

Wake-Up Call

In fact, Coke—which placed a close second to Unilever in Ceres’s ranking of companies with the best water practices—learned that lesson the hard way in 2004, when it faced consumer backlash and had to close a plant in India over accusations (since rejected by a court) that the facility was siphoning water from a local community during a drought.

Coke Chairman (and former CEO) Muhtar Kent labeled that event a “wake-up call” that pushed the company to make water management a priority and cut its consumption. Last year, Coke announced that it was on track to meet its goal of reducing use of water in manufacturing by 25 percent by 2020, using 2010 as a baseline. It has already reduced the amount of water needed to make one liter of soda to 1.98 liters, which means 0.98 liters are used in the manufacturing process. By 2020, it expects to bring the total down to 1.7 liters.

How is Coca-Cola accomplishing this? Well, unlike the recipe used to make its main product, there is no secret formula involved. Indeed, it mainly relies on off-the-shelf technologies that come from outside suppliers—technologies that are also in wide use among other major beverage makers. There’s even a 20-member technology- and best-practices-sharing trade group, the Beverage Industry Environmental Roundtable (BIER), that Coke cofounded in 2006 “to advance environmental sustainability within the beverage sector.”

Nevertheless, Coke itself doesn’t have a one-size-fits-all approach to water conservation. That’s because it’s also a heavily franchised operation. The Atlanta-based Coca-Cola Co. makes the syrups, bases, and concentrates used in its drinks, and handles brand management and marketing. But the manufacturing, bottling, and distribution of its products is handled by its more than 250 bottling companies, all franchisees, at more than 900 plants. The parent company sets the water-use standards, and it’s up to each bottling franchise to determine how best to meet them in its plants. Coke doesn’t order its bottlers to use any specific type of equipment. It may recommend some technologies, Bowen says, but will then tell them: “You have to figure out what works best for you within your budget, within the configuration of your plant, within your personnel structure.”

Vulnerability Map

To deal with scarcity, the first thing Coke considers is location, says Bowen, who has a doctorate in environmental systems engineering. “We really try to avoid siting plants where we know there is a water issue.” As part of that effort, Coke used sophisticated hydrological modeling to create a global map of water risk that predicts how vulnerable areas will be up to 50 years into the future. It’s since donated the map to the World Resources Institute, where it’s publicly available online.

But Coke can’t always avoid setting up plants in water-scarce areas. For instance, it opened a bottling plant last year in the Gaza Strip, which endures frequent water-treatment and supply crises. The Democratic Republic of the Congo, where access to clean drinking water is reported to be among the lowest in sub-Saharan Africa, nonetheless has a Coca-Cola-licensed bottler of Dasani water. If Coke determines it has no alternative but to place a plant in a region where water supplies are tight, “we try very much to find sources of water that the communities don’t depend on,” says Bowen. Richard Crowther, a mechanical engineer who was Coke’s former director of environment and sustainability and is now a senior consultant for environmental consultants Antea Group USA, says aquifers are typically stacked, and communities in developing countries usually use the ones closest to the surface. That gives Coke space to drill more deeply, often thousands of feet, to find separate sources that don’t cross-feed into the community aquifers. For each of its plants, Coke does a Source Vulnerability Assessment every three to five years that looks at a number of issues, including susceptibility to drought, potential for contamination, pricing structure, and legal requirements. Each plant must then submit a Source Water Protection Plan that helps it devise efficiency, water treatment, and wastewater treatment programs.

Returning the Water

Those plans are also instrumental to Coke’s replenishment program. Last year, the company announced that in 2015 it not only met but exceeded its goal of returning to communities or nature an amount of water equal to the amount of product it produced. Between 2014 and 2015, it returned—either directly to sources used to manufacture drinks, or to areas outside those watersheds—some 50 billion gallons of water, or 115 percent of the water contained in its beverages that year, mainly via efforts like reforestation and reestablishing wetlands. Still, the whole soft-drink manufacturing process uses more water than is returned.

A Dry Rinse

To cut water usage in its bottling plants, Coke focuses on three steps: efficiency, technology, and production process design.

Efficiency can include a variety of things. Identifying and fixing leaks, for instance, can go a long way toward cutting water use. The conveyor belts in plants are metal, and many plants have switched from soap-and-water-based lubricants to dry lubes that are silicon-based and use no water. Some plants now use air instead of water for rinsing packaging. And Bowen has worked with manufacturers of fillers, the machines that fill cans and bottles, to design them so they can be rinsed with less water. Also, he says, bottlers often reuse some of the water used to flush out mixing tanks. “The last flush is typically the cleanest flush. They’ll reclaim that water and reuse it as the first flush in subsequent cleaning.”

Much of the technology used in plants is for treating water—both incoming water and wastewater. All water flowing into a plant is treated by the bottler, even potable water from dependable municipal sources. That’s because the water used to make drinks has to be as pure as possible to avoid altering the flavor. According to Crowther, the rule of thumb for all beverage plants is that, after treating incoming water, 80 percent will be usable for products and 20 percent will be rejected because it’s full of concentrated contaminants. (That ratio can range from 90/10 for plants receiving highly treated water to 70/30 for plants whose sources haven’t been pretreated.) The rejected water, however, can be further treated for non-beverage uses, including cleaning, feeding cooling towers, and flushing toilets. “You want to recycle it as many times as possible,” Crowther says. If the wastewater that is eventually pumped out is being returned directly to nature, the bottling plant will also treat it one last time. “It has to be clean enough that it can support marine life,” he adds.

The technologies bottlers use to treat water vary. “Every plant treats their water somewhat differently,” Bowen says. “There is what we call a multiple barrier process, multiple barriers to contaminants.” These typically involve reverse osmosis, activated carbon filters, and chemical treatments. In the U.K., a bottling plant in Wakefield, West Yorkshire, recently installed a membrane ultrafiltration system, a pressure-driven purification process that gave it a tenfold reduction in wastewater generation. It was developed by Norit Membrane Technology, a Dutch company.

Israeli company Atlantium Technologies has developed a chemical-free water disinfection system that uses ultraviolet light, fiber optics, and hydraulics. It’s now in use in around 250 Coke plants. Rotem Arad, an executive vice president at Atlantium, says fiber optics enable the sending of UV light long distances by reflecting it and using it over and over again. Traditional UV water treatments use light from one source, but that can create “a shadow effect” that allows some microbes to survive. “In our technology, basically nothing can escape.” The system can be used to take out pesticides and bacteria, but also to remove chlorine. Plants often use chlorine to disinfect water, which then must be dechlorinated before making sodas. “Otherwise it would taste terrible,” Arad says. Wastewater returned to nature also must be chlorine free.

Another Israeli company, Blue I Technologies, makes advanced controllers and analyzers for water treatment and also has worked with many Coca-Cola bottlers. For bottlers, says Amit Shilony, the company’s vice president for North America, the focus is ensuring that the water is chlorine free, “which is interesting, because it’s harder than measuring how much chlorine is in the water.” The Blue I system can give early alerts to bottlers if it detects a problem, well in advance of the water being used in a product.

Coke rarely develops its own technologies, Bowen says, because that’s not a core aspect of its business. Moreover, he adds, Coke’s reliance on proven, off-the-shelf technologies just makes good business sense because “we have a brand to protect.” Or as Crowther says, “you can’t afford to be someone’s test bed.”

The third and most complex—but important—step in cutting water use in beverage plants, Crowther says, is production process design. “That’s where the real opportunities exist” for savings. And that, essentially, comes down to industrial engineering.

For example, says Serena Levy, Coca-Cola’s communications director, “we’ve also been innovative with how long you can run certain products on the line. Every time you change the line out, you’re not only losing productivity; you’re using water [for cleaning]. So, can we run all the Coca-Cola we’re going to need for the entire month on one line run? Then you gain some productivity on water use.” And the complexity of that task is head-spinning, Bowen says. Many plants have five lines, each doing 600 to 800 cans or bottles a minute, all different flavors. “And every time you change flavors, you have to clean, rinse, sanitize, and cool down every piece of equipment in that line. The production manager does not have an easy job.”

Coca-Cola’s efforts to shrink its water footprint are, Bowen says, driven by both a desire to be a good corporate citizen and demands placed on its business by water scarcity. “It’s something we recognize as an important issue. It’s important not only to us, but to the communities where we operate. There is a social license to operate, and there’s a business license. You have to maintain both of those.”


By Thomas K. Grose

Thomas K. Grose is Prism’s chief correspondent, based in the United Kingdom.
Design by Michelle Bersabal

Mentor, Builder, Partner

Bevlee Watford’s advice to a struggling student sums up her mission as an educator and activist on behalf of underrepresented minorities: ‘Work with me. You can make it.’

When becoming a mining engineer meant being the only African-American in a department of about 100, Bevlee Watford was ready. When paying for school meant working summers and winters—at one point shoveling coal in Pennsylvania—she did it. When her passion for counseling students—particularly women and minorities—meant abandoning a tenure-track job, taking a pay cut, and starting a new program from scratch, she jumped at the chance. And when a
Supreme Court ruling threatened the program’s survival, she saved and even expanded it.

The career of ASEE’s 2017–18 president, the first female African-American to hold the title, offers multiple examples of seizing unfamiliar challenges, overcoming obstacles, and building durable institutions with the capacity to change lives. Along the way, this no-nonsense trailblazer has made a lasting impact on students at Virginia Tech, where she is a professor of engineering education and associate dean for academic affairs, and become a leader of—and role model for—champions of diversity in ASEE and beyond.

If success in engineering for women, blacks, Hispanics, LGBT students, and those with disabilities is a matter of social justice, Watford has an additional take. She’s fond of quoting Bill Wulf, former president of the National Academy of Engineering, who contended: “Diversity is essential to good engineering!” Varied life experiences of men, women, people from different ethnic backgrounds, and the handicapped, he argued, represent “the gene pool out of which creativity comes, out of which elegant engineering solutions come.” Within ASEE, Watford grew frustrated when nothing came of a 2002 diversity plan. When she later chaired a Society Diversity Task Force, she made sure that it resulted in a strategic plan, a standing committee, and a website.


As a longtime Hokie, Watford smiles at the memory that when her high school guidance counselor suggested that she apply to Virginia Tech in Blacksburg, then known as Virginia Polytechnic Institute, she had never heard of the school. One of its draws was cost: It was cheaper even for an out-of-state resident than going to one of the State University of New York schools in her home state. Watford knew she would have to pay her own way, so she established a routine: attending the fall semester, working the winter, returning to school in the spring, and going back to work in the summer. That included shoveling coal in a mine south of Pittsburgh. The stint fit nicely with her undergrad major in mining engineering, where she was the only African-American student and one of four women out of a department of roughly 100. “It was very hard labor,” she recalls, but fellow mineworkers “didn’t make us do any more than we could handle.”

Graduating in 1981, Watford contemplated doing an M.B.A., but a faculty member urged her to consider a graduate degree in industrial engineering and operations research. She met Prof. Timothy Greene, who had just received a grant to study coal transportation from West Virginia to Norfolk, Va. Trouble was, Greene didn’t know anything about coal—“not a dang thing”—so he hired Watford and served as her adviser for both her master’s and Ph.D. She was the first black female from the College of Engineering to earn advanced degrees in those fields.

Blacksburg Beckons

Accepting a tenure-track job at Clemson, Watford soon found that she did not like doing industrial engineering research and publications. “What I really enjoyed,” she recalls, “was working with students and helping them succeed. Just being there, being someone students could come and talk to and share their issues and concerns with and see me as a role model made me feel that I could make a real difference in their lives.” Her first National Science Foundation grant at Clemson created a summer program introducing high school students to engineering. She also became friends with Sue Lasser, now an academic enhancement counselor at Clemson, who founded and directed Programs for Educational Enrichment and Retention (PEER) to reach out to underrepresented minorities. Greene, meanwhile, hoped to entice Watford back to her alma mater, tailoring a position of director of Minority Engineering Programs at Virginia Tech to fit her background and abilities. Watford grabbed the opportunity. “That was exactly what I wanted to do.” She did so despite a warning from the president of Clemson that she was making a “tragic career move”—and despite taking a pay cut. It was, Greene recalls, “one of the best decisions Virginia Tech ever made.”

Besides support from Virginia Tech colleagues for her new venture, Watford says, “I gained a lot of knowledge from the professional community, particularly the National Association of Multicultural Program Advocates (NAMEPA) and the Women in Engineering ProActive Network (WEPAN). These are professional organizations of people doing the job I was tasked with doing. They helped me understand best practices and how to build a program from scratch.”

Watford began with a peer-mentoring program for black students before expanding her initiative to include similar programs for women and Hispanics. She is particularly proud of Hypatia, a living-learning community of 300 women who live and work together in engineering. In the first year after it was created in 2001, the graduation rate of female engineering student participants increased by 20 percent.

She also spearheaded Imagination, a one-week math and summer camp for seventh and eighth graders to do hands-on activities with science and technology; a Pre-College Initiative (a collaboration with Virginia Tech’s National Society of Black Engineers chapter) that includes five daylong events to help students prepare for the college application process; and Women’s Preview Weekend, which enables women who have been offered admission to the College of Engineering to spend a weekend staying in residence and meeting people from the community.

‘A Game-Changer’

One student who quickly came to appreciate Watford was Tres Wooldridge, an African-American mechanical engineering student who entered Virginia Tech the year before Watford returned to the school in 1992. “At the time, I was a student who was really struggling,” Wooldridge remembers. “I had a tough time adjusting to university. I didn’t have a network that could help me get through college. I thought of quitting.” He adds, “When I first met Dr. Watford, I was complaining about how bad the system was set up and I didn’t think I could make it. She pulled me into her office and said, ‘Work with me. You can make it.’ That was a game-changer for me. She started putting programs in place and putting me in touch with mentors who could help me. I can guarantee you that was one of the main reasons I ever graduated.”

That inspired Wooldridge to become a mentor himself in one of
Watford’s programs. “I saw what Virginia Tech was like before she showed up and I saw what it was like after she showed up, and I had to be a part of it. Mentors for incoming freshmen, summer camps—I was all into it, man. I was able to take the knowledge I had and impart it to others. That not only provided a sense of fulfillment; it also tapped into leadership skills I was able to use later on down the line. I couldn’t have known that at the time, but it would have a big effect on my life.” Watford put Wooldridge in touch with a contact at Corning for an internship. “Today, I’m in my 21st year working at Corning as a plant manager of cable components in Phoenix. It wouldn’t have happened without Dr. Watford, and I bet you many other students have similar stories about her.”

The whole program was almost dismantled following the Supreme Court’s 2003 decision in Grutter v. Bollinger, launched by a white student who contended she had been denied admission to the University of Michigan Law School because of race. While the court sided with Michigan, the ruling prompted institutions to rethink programs aimed solely at minority students. Virginia Tech rallied, expanding the program to include every student and renaming it the Center for the Enhancement of Engineering Diversity (CEED). Now, instead of supporting 350 to 400 students, CEED caters to a majority of the entire freshman class of more than 1,300 students.

Richard Benson, president of the University of Texas at Dallas who was Virginia Tech’s engineering dean from 2005 to 2016, found out firsthand about the inclusive nature of CEED. “I’m a white male with a white male son who just graduated from Virginia Tech this May. Like a lot of new students, he found engineering tough and he needed help—and he got it from some of the superb programs that Bev has put into place. I was already a fan, but it was pretty powerful to revisit those programs through the eyes of a father.” He adds, “Bev is absolutely the best in making sure that students achieve success. Her son, Devon, graduated from Virginia Tech and daughter, Leah, next year, and Bev has almost a mother’s interest in kids doing well. She has had a colossal impact on student success at Virginia Tech.” Just before he left Virginia Tech, Benson was instrumental in Watford’s receiving the school’s 2016 Presidential Principles of Community Award.

In 1997, Watford took on a second job at Virginia Tech: associate dean of engineering for academic affairs, which she says involves “everything from recruitment and ABET to teachers and curriculum.” This is someone who Sue Lasser says “crackles with energy—a really positive person.” Benson calls her “a real dynamo—someone who works incredibly hard and incredibly well.”

A Wide Swath

While veering away from the research-and-publish grind early in her career, Watford found a niche in scholarship by collecting data and documenting results of her Virginia Tech initiatives and pulling in millions of dollars in external funding. She twice served as a NSF program director, most recently of the Broadening Participation program within the Division of Engineering Education and Centers, and is profiled along with other “pioneers” of engineering education research by the University of Washington’s Center for Engineering Teaching and Learning.

Louis Martin-Vega, who worked closely with Watford when he was president of ASEE and she was president-elect, says that he has seen up close the range of what Watford has accomplished. “The world of engineering education, the world of increasing participation of underrepresented minorities, the world of women in engineering—Bev cuts a wide swath. She has tremendous determination and consistency and has followed the beat of her own drum.”

Martin-Vega describes her election as a “real milestone. There are people out there who will be very proud of the fact” that ASEE has elected its first African-American female president. “It is meaningful to a whole segment of the population—people who can point to that and say that there has been a breakthrough and I too can aspire to achieve something great.”

Among Watford’s goals as president, as spelled out in her Candidate’s Statement, is to build on ASEE’s relationships. “I see the president as having a key role in enhancing partnerships with strategic organizations to achieve real and lasting change,” she wrote. On receiving the gavel, she announced that ASEE will co-host the First Annual Conference of the Collaborative for Inclusion and Diversity in Engineering and Information Technology—a conference intended to bring together all those working to achieve increased participation for underrepresented groups. This promises to be an active year.


By Pierre Home-Douglas

Freelance writer Pierre Home-Douglas is a regular contributor to Prism.
Design by Francis Igot

Great AI Leap Forward


Challenging U.S. leadership, Chinese industry, academia, and consumers are making artificial intelligence a way of life.

Inside one of this city’s busiest subway stations, throngs of commuters walk down a long passageway as they transfer between subway lines. The walls and columns are plastered over completely with bright green pop-art-inspired billboard advertisements, impossible to ignore. “Wow!” a white-collar worker exclaims in one ad, gushing about the virtual tutor on a language-learning platform powered by artificial intelligence. “My AI teacher caters so well to my needs, I’m obsessed with learning English now.” The marketing campaign by the start-up Liulishuo seems to be working. Introduced in July 2016, the company’s “AI English teacher” app boasted almost 600,000 paying users a year later. It’s one example of how mainstream AI has become in China. Another comes from a computer scientist, speaking at a recent academic forum here: Even taxi drivers, he says, want to discuss deep learning and AI with him.

From major tech and Internet companies to start-ups, airports, local police, hospitals, and individual consumers, Chinese are rapidly developing and using what Google, which itself has a big stake in AI, defines as “the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.” Given China’s growing investment in research and AI-oriented start-ups, including some in the United States, this expansion should only continue. In July, China’s top-level State Council put its official stamp on the trend, declaring that by 2030, the country will become a leading (one translation said “world’s premier”) artificial intelligence innovation center. “We must take the initiative to firmly grasp the next stage of AI development to create a new competitive advantage, open the development of new industries, and improve the protection of national security,” the council said.

Liulishuo (that’s Mandarin for “speak fluently”) draws on a combination of speech recognition, machine learning, and big data—along with gaming, social media, and neuroscience—to create a personalized learning experience: the customer-friendly “AI English teacher.” Founded in 2012, the firm is led by CEO Wang Yi, a Princeton computer science graduate who spent two years as a Google product manager before he and his co-founders seized the chance to return to China and build a company using AI technology. An obvious niche was the high demand among ambitious Chinese for English-language skills. Wang thought he could improve on the English-language training schools doing a brisk business in China. As Bloomberg has reported, Liulishuo “built algorithms that quantify dozens of dimensions of speech,” such as pronunciation, grammar, and fluency. Rather than requiring memorization, the formulas are designed to activate subconscious pattern recognition. Entertaining app users as they learn, Liulishuo includes a feature that lets them dub their voices over movies and TV shows. Launched in 2012, Liulishuo’s free version has drawn over 45 million users. The company brags it can teach English in a third of the time it would take with a human teacher in a training center. In turn, data from these users let the firm generate what it claims is the world’s largest speech bank of Chinese people speaking English and refine its methods of adapting to individual learners.

Surf that Wave

At 39, Wang is the elder statesman of his company, but he looks as youthful as his millennial colleagues busily working on a mobile English learning platform at the company’s Shanghai headquarters. With summer temperatures outside soaring above 100 degrees, he’s dressed casually in black geek-chic glasses, shorts, and a gray T-shirt that reads, “I love it when you call me Big Data.” Chinese entrepreneurs, says Wang, “know that they have a big opportunity in front of them, and they are trying their best to capture that. The wave is half-formed and for any good surfer, you just want to surf that wave.”

China aims to be the crest of the wave. While some say Beijing came late to the AI game, it wasn’t standing still before July’s policy announcement, which set a target of $59 billion in output by 2025 focusing on AI software and hardware, intelligent robotics and vehicles, virtual reality, and augmented reality R&D. In February, Chinese Internet giant Baidu announced it had received funding from China’s National Development and Reform Commission to lead the creation of a national deep learning lab. In March, Chinese Premier Li Keqiang mentioned AI for the first time in his annual Work Report as part of the National People’s Congress, thereby flagging AI as a priority.

A Go tournament in May in the city of Wuzhen became “a sort of Sputnik moment” for Beijing, spurring July’s announcement, the New York Times reported. AlphaGo, a machine built at the Google-owned DeepMind artificial intelligence lab, beat Chinese grandmaster Ke Jie, the world Go champion, in a three-game match. China’s government blocked live coverage, but the tournament nonetheless made an impact. Wang recalls: “By creating the drama of human Go champions playing with AI, it really caught people’s attention, and it demonstrated, although arguably in a very restricted setting, [AI] is good enough.” China may in fact not have been far behind Google in the AI-Go sweepstakes. In March, Fine Art, a program created by the Chinese company Tencent, won the 10th Computer Go UEC Cup championship, competing against 30 of the world’s best AI Go programs, such as Facebook’s Darkforest.

Well before the State Council announcement, Chinese academics were deep into AI, churning out papers in the discipline at twice the rate of their U.S. counterparts between 2010 and 2015 (although they ranked only 34th in terms of citation impact). In the same time period, the number of AI patent submissions coming from China jumped 190 percent while U.S. patent submissions rose 26 percent (but to a total of more than 15,000, much more than China’s 8,410). “Chinese research in AI has grown exponentially, unseating the early leader, the United States,” concludes the U.S.-China Economic and Security Review Commission, created by Congress. Meanwhile, Chinese companies are “ringing alarm bells” in Washington by investing in cutting-edge U.S. start-ups, the New York Times reports. Citing an unpublished Pentagon-commissioned white paper, the newspaper says: “Beijing is encouraging Chinese companies with close government ties to invest in American start-ups specializing in critical technologies like artificial intelligence and robots to advance China’s military capacity as well as its economy.”

Before Boarding, a Face Scan

So far, industry has been the most visible driver of the AI revolution in China, particularly with facial-recognition technology. Some airports and tourist sites have already put face recognition to use for admission and check-in and boarding. The South China Morning Post reports that China Southern Airlines began allowing passengers at an airport in Henan Province to have their faces scanned in lieu of using boarding passes. Alibaba announced earlier this year that face recognition technology may be incorporated for mobile payments. Increasingly, local authorities are using AI to control behavior. Shanghai’s police recently announced they are the latest of several cities’ police departments to roll out face-recognition technology to catch and shame jaywalkers. Chinese companies are developing technology that they claim can identify and catch suspects before they even commit a crime. And last year, Beijing-headquartered Baidu, one of the world’s biggest Internet companies, made its cross-age face recognition technology available to a nongovernmental organization helping families find trafficked children. The technology, also known as age-variant face recognition, distinguishes between facial features that change with age and those that remain intact to identify people based on images captured when they were younger. So far it has been successful in reuniting several families, including a family whose son was abducted 27 years ago. Voice-recognition software is also spreading. In July, Alibaba launched Tmall Genie, a digital assistant speaker similar to Amazon’s Echo, which performs tasks and delivers information in response to voice commands.

Baidu’s autonomous car project is attracting a lot of attention for the potential applications it could have in public transport. Baidu CEO Robin Li thinks that a driverless car will be on the road in China by 2020. That’s assuming China’s legal framework catches up. After he demonstrated Baidu’s autonomous car through livestreaming at a Baidu Developer Conference earlier this year, Beijing’s traffic police rebuked him for breaking a law requiring drivers to keep their hands on the steering wheel.

AI advances are also being applied in health care. Some Chinese hospitals use AI to read CT scans. Start-ups are using AI to help patients keep track of and manage their personal health. Last year Baidu unveiled a chat bot for patients and doctors to communicate. This spring a start-up unveiled an AI computer vision tool to diagnose lung cancer. Another new firm, A-Tag Technology, is investigating the use of face-recognition technology to identify skin diseases and rashes.

Big Market = Big Data

Academics and entrepreneurs cite several reasons to be bullish on China’s future as an AI superpower. China’s massive population, which can be seen as a disadvantage in urban livability, provides not only a vast potential market for AI technology but a source of data for such applications as voice and face recognition. “Big market can also be understood as big data. China has so many people, so that means China has so much data. That means AI systems can be trained very well and work very well,” says A-Tag Technology cofounder Zhang Haoxi, an associate professor at Chengdu University of Information Technology who works on machine learning and the Internet of Things.

“If you have a small market, you don’t necessarily need to use AI. When you reach a certain level of scale data set, then AI becomes really meaningful,” says Wang. While Chinese research collaboration with academics in the United States and elsewhere has grown in recent years, the amount of data available in China adds to the international appeal, says Feng Jian Feng, a professor of brain science at Fudan University’s Institute of Science and Technology for Brain-Inspired Intelligence. “It’s really for mutual benefit.”

A June report published by Washington’s Woodrow Wilson International Center for Scholars noted that cutting-edge AI technologies are often deployed faster in China than in the U.S., due to high customer loyalty and speedy integration of AI into existing products. Baidu’s online translation system, for instance, used neural networks a year before Google’s neural machine translation caught industry attention.

Another benefit to AI technology researchers and developers stems from the country’s tight security apparatus, which has accustomed the population to more government intrusion in their lives than most democracies would allow. As a result, privacy protections are weaker than in the West, and Chinese society in general is used to private data being captured without an individual’s permission. “In the U.S., it is not easy to acquire so much data. In this way, we have an advantage,” says Gao Shenghua, an assistant professor of computer vision at ShanghaiTech University. Zhang Haoxi adds, “In the U.S. the rules are very settled. You cannot break the laws or rules, and sometimes this can be a disadvantage to trying new things.” In China, he says, “We have a very flexible environment for trying new ideas. Sometimes there are no clear rules for something. Then it becomes an advantage for start-ups to do business. It gives them windows to try new ideas.”

‘Not a Top Research Country’

If entrepreneurs feel less restrained, the same can’t necessarily be said of academic researchers, whom China needs to pursue the discoveries that form the fundamental building blocks for innovation. Low pay, lack of a tenure track, and a backward system of research funding loom as potential impediments to China’s drive to become preeminent in AI. Dawn Song, a professor of electrical engineering and computer science at the University of California-Berkeley, contends that China still lacks an environment where researchers can thrive. “China is not a top research country yet,” Song, a 2010 MacArthur Fellow, told a recent forum in Shanghai entitled Securing the Future with AI. “We need to have a good ecosystem. If you want to attract the best people, you have to create the best environment and retain the best talents.” She added, “For now, they export the best students to all go study abroad.”

Li Kai, professor of computer science at Princeton, who spoke at the same forum, agrees. “There has to be a culture to support you to do whatever is important,” he says. Having earned bachelor’s and master’s degrees in China and then a Ph.D. from Yale, he is familiar with both U.S. and Chinese systems and says the research environment here is not yet one that allows academics to freely pursue their interests. “It can’t be too linked to funding and government-led initiatives. I know many professors in China agree with me. The scientists are not brave enough or capable enough to work on what is most important to them. They still have to find the funds for their research.”

Many students eventually return to China, but when they do, there’s a good chance they’ll be lured away from academia by industry, which can pay three to five times as much. “The market demand for AI talents has been surging in recent years, so there’s a shortage at this time,” says Liulishuo’s Wang Yi. “I think in China, because there’s a huge supply of venture money and a shortage of AI talents, this may be more pronounced.” Government funding doesn’t match its aspirations, some academics say. The grant system, while getting stronger, “is less forward looking and less flexible” than is needed, says Zhang Zheng, professor of computer science at NYU Shanghai. “The grant money is not up to the U.S. standard. The hot money is all in the VC [world]. Why not divert some of that to basic research that doesn’t [yet] exist?”

China also lags in partnerships between Chinese industry and universities, some academics say. Zhang Haoxi wants Chinese universities to forge more relationships with companies of the kind that allows Yann LeCun, Facebook’s AI Research director, to retain a part-time professorship at New York University: “This could be a very good thing both for universities and industry development for applications.” Some schools are headed in this direction. One is ShanghaiTech University, which opened in 2013 in Shanghai’s Zhangjiang Hi-Tech Zone, surrounded by R&D facilities, and has adopted a tenure track. Gao Shenghua, an object-and-face recognition expert who joined the university after a postdoc in Singapore, says the school encourages industry cooperation. His department, the School of Information Science and Technology, is collaborating with United Imaging, a nearby company offering medical solutions using computer imaging.

Research Vs. “Quick Money”

“How we can get businesses to be involved in long-time research? This is a big problem,” says NYU Shanghai’s Zhang Zheng. He added, “There is no long-range research yet in the VC community here. They are near-sighted.” Zhang Haoxi added, “To have our own companies which are willing to invest in real research rather than making quick money, this is basically the main challenge for China.” He said that to some extent Alibaba and Baidu are doing this, but not enough. “There is still a long way to go for them.”

Sufficient capital isn’t a problem, says Wang Yi. “Some of the AI companies in China have even higher valuations than U.S. companies these days. People are saying there’s no shortage of venture capital; there’s just shortage of good companies to invest in.” But Zhang Zheng says investments are narrowly focused. “In China we have a lot of hot money focused in one or two areas, spent in a limited area.”

Whatever gaps exist now in China’s investment strategy, the country has a track record of rapidly catching on and catching up. Can cutthroat U.S.-China competition be avoided? Alibaba chairman Jack Ma sought to play down that prospect at a recent World Intelligence Congress in Tianjin, saying, “We need to put more time [into] thinking about the future, not about how to fight our competitors. We need to work together with the U.S. to solve challenges. We have already passed the era of competing.” The U.S.-China Economic and Security Review Commission has a different take: “Competition between U.S. and Chinese firms is only set to increase as the Chinese government is throwing its support behind its commercial and military AI researchers.” Not long ago, Chinese entrepreneurs were infamous for snatching and exploiting U.S.-grown technology. That’s changing. “The U.S. entrepreneurs are now watching more closely what’s happening in China and wanting to copy or get inspired by what’s happening here,” says Yi. If he’s right, U.S. billboards could soon show delighted American students praising their “AI Chinese teacher.”


By Rebecca Kanthor

Rebecca Kanthor is a Shanghai-based writer and a correspondent for Plastics News. She works part-time as a proofreader for ShanghaiTech University, one of the institutions mentioned in this article.
Design by Nicola Nittoli