AI is everywhere: inside smartphones, on wearable devices, at hospitals and airports, and even in cars. The technology is shaking up economies, pushing companies to adapt and forcing governments to take a closer look at its implications. As AI becomes more ubiquitous, it is also polarizing public opinion, with fears over man’s ability to control the technology, its long-term impact on labour markets and concerns over data collection and privacy.
The late physicist and author Stephen Hawking warned that AI could be the “worst event in the history of our civilization.” Similarly alarmist, Elon Musk, founder of electric car manufacturer Tesla described AI as more dangerous than nuclear weapons.
Improper use of data risks tarnishing AI industry development, which hurts everyone in the long term argues Ericson Chan, CEO of Ping An Technology, part of China's Ping An Insurance Group, which was listed as the world's largest insurer in the 2017 Forbes Global 2000 list of most powerful companies.
The company invests 1 percent of its revenues into technology development and its leaders are passionate advocates of AI as a force for good in society. To develop AI in a responsible way, Chan says, “it is critical that you have the proper governing approach and make sure data are all anonymized.”
The philosophy behind Ping An Technology’s investment in AI, according to Chan, is to “change people’s lives for the better in many ways and forms.” Most critically, AI has democratized many services and “commoditized a lot of privileges” that were previously unavailable for large sections of society. AI also has the capability to predict and prevent many kinds of risks, thus improving day-to-day life, he adds. “The net result of all this is increased efficiency in many things we do,” Chan observes.
Fundamental services such as finance, education, healthcare and security are where Ping An focuses its research and development. Within finance, robo-advisory is gaining traction. “A lot of privileged services that are available to private banking or high net worth customers today, such as advisory services, can be provided to a wider range of customers using AI,” he says.
Similarly, AI offers the opportunity for underserved segments of the population to gain access to finance even if they lack a credit history or proof of address. By using AI to understand patterns based on the apps people use, the credit team can “make some judgments that are probably more accurate than a utility bill to know where you live.” It’s good for the company as well as the customer, argues Chan, since knowing the customer better lowers the risk of credit loss.
In education, AI programs can improve student outcomes dramatically. Noting that every student is unique, Chan adds that an AI can be used in student homework programs to respond and adjust to their learning needs, resulting in a tailor-made curriculum. It can also, using facial recognition, give certainty to education authorities that the correct students are sitting exam papers.
Facial recognition technology is also finding application with security services. AI driven programs allow for real-time face matching with a database comprising thousands of catalogued photographs. The company’s publicly shared use case is at Shenzhen airport where AI models increase public safety by cross-reference millions of faces with the authority’s database.
As an insurance company, health and protection have always been themes that are core to Ping An’s mission. In recent years this has translated into heavy investment into biomedical technology and services.
There are four different healthcare applications that Ping An is focusing on its development of AI. These are to predict the outbreaks of disease, diagnose conditions, assist doctors, and provide quality control for a variety of hospital services, says Ping An Technology's deputy general manager, Geoff Kau.
By applying artificial intelligence to huge amounts of patient and claims data, Ping An can forecast the spread of contagious diseases, giving municipal governments time to prepare for major outbreaks and allocate resources within time. Ping An is working with the Chongqing municipality in southwestern China to forecast the spread of different flu strains and other contagious diseases a week in advance, with over 90 percent accuracy.
The company also uses AI models to diagnose complex diseases. For example, it can look at a CT scan of a lung and identify small masses of tissue against an existing database to predict if there is a potential for cancerous growth, Kau explains. The company’s lung nodule detection model was ranked first in the LUNA16 Grand Challenge medical imaging research competition, where its deep learning approach set a world record for accuracy.
Ping An’s AI team is broadening the number of health issues that they focus on. This includes looking at the relation of bone density to fractures and identifying 14 different lung complications based just on X-ray imagery, which is often the main imaging technology available in lower-tier cities and rural areas of China. “The disease types will be broadened, and we will go down the value chain,” says Kau, adding that the team “estimated that if we had disease models covering the top 25 diseases nationwide, 400 million people would be covered.”
With doctors in many Chinese hospitals seeing as many as 200 patients each day, time is a precious commodity. AI is being used to collect advance information from patients to do basic pre-diagnoses, leading to shorter and more targeted consultations. AI-equipped virtual robots can also help with post-hospitalization follow-up by monitoring patients’ blood sugar and other health indicators. “If new symptoms develop they can ask further questions and flag a patient as high risk and have the doctor follow up,” Kau adds.
Where quality control on imaging and clinical treatments is ordinarily conducted by hospital inspectors at with small sample sizes and at random, AI can provide a methodological approach. AI models installed at hospitals can scan every diagnostic image to validate image quality and clarity (for which the patient needs to be correctly positioned) so that the result can be read with confidence and radiologists given further training where necessary.
According to The Lancet, only half of China’s three million doctors have bachelor degree-level education. This creates a huge imbalance in healthcare quality across the country, which AI is well-placed to address.
There can be no question that responsibly deployed intelligent machines can transform industries and ecosystems for the better. Yet those who oppose the unfettered growth of AI still worry about its potential misuse.
Self-regulation by the industry using a “moral and ethical compass” is the way forward, says Kau. He notes that governments also need to play a bigger part in ensuring the ethical use of AI.
For Chan, the bigger question is: “As AI brings about change and disruption, is society prepared for that? The governments, the education [systems] and society have a role in preparing people for this new world.”
He notes a lot of jobs will go away thanks to AI but they would be replaced with new jobs. “Kids who are studying today, by the time they graduate, will be taking up new jobs that we don’t even know about today,” says Chan.
Society will ultimately decide whether it will use AI as a force of good or evil, not the technologies themselves.