Governments across the world have turned to technologies such as artificial intelligence and facial recognition to provide assistance in the battle against the coronavirus. There are still many ethical, privacy, and security concerns.
KrASIA had a discussion with Fanglin Wang, head of computer vision at Singapore-based big data and AI company Advance.AI. Founded in 2015, Advance.AI uses artificial intelligence to reduce fraud risks and improve operational efficiency for their clients, serving operators in sectors such as fintech, banking, and retail. Last year, Advance.AI raised USD 80 million Series C for its regional expansion in Southeast Asia.
In the interview, Wang called for governments to ensure the ethical and sustainable use of facial recognition technologies and AI, including during the current coronavirus crisis. As for businesses, they will have to implement measures to trim inefficiencies after the pandemic subsides.
KrASIA (Kr): The COVID-19 pandemic has put technology powered by artificial intelligence in the spotlight, including facial recognition. Coronavirus-related surveillance has been used widely in hard-hit countries such as China or South Korea. What are some of the concerns arising from this trend?
Fanglin Wang (FW): Although facial technology has been around for approximately ten years, we are still at the beginning of a long journey. Remember, facial recognition technology is already being used in daily life. Mobile banking applications using FaceID for authentication are already very common. When people upload selfies to social media platforms, they are indirectly powering the world’s largest facial recognition database. When social media automatically tags you in photos, that’s facial recognition technology at play.
We have seen some practical applications of facial recognition technology during the coronavirus outbreak. The first example is around security: building identity and temperature checks, as well as entrance registration at buildings. All of this can be done online and remotely, ahead of the building security team, to speed up the process while minimizing human contact and spread of the virus.
When human interaction and contact must be limited, such as during the SARS outbreak nearly two decades ago and the coronavirus pandemic today, China saw a huge spike in demand for fresh fruits and vegetables on e-commerce platforms. The only way for merchants and suppliers to get on these platforms and rapidly scale to meet demand without face-to-face interaction was through facial recognition, digital onboarding capabilities, and ID authentication technology.
Certainly, there are a variety of legitimate concerns. This first is around privacy and the question of whether I am being constantly monitored. The second is around how data is being stored, collected, and used. Linked to this is security and how securely the data is being stored. There is also the question of accuracy. AI algorithms have to be extensively trained and tested by relevant data to be accurate. For example, local data of Southeast Asian faces and facial structures, skin tone, and so on are required for high accuracy of facial recognition technology with this population.
Every country needs its own regulation and enforcement to prevent bad actors and abuse. People’s right to privacy and how their data is being used must always be a top priority. For example, data must only be used for specific and stringent purposes, such as for law enforcement, but not necessarily broad and unrestricted surveillance.
It’s very important for governments and companies to be clear about how facial recognition technology will be used. I also think various countries are at different stages of their AI journeys. In China, where facial recognition technology is an accepted part of life, people have come to accept it. In the US, by contrast, the right to privacy and civil liberty is enshrined by law and can clash with facial recognition applications in public and private life.
Kr: There are so many ethical, privacy, and security concerns surrounding facial technology. What should be the priorities that governments have to address in terms of regulation?
FW: Right and ethical use of AI is a constant learning process and needs to be debated openly. This is why Singapore proposed updates to its AI governance framework at Davos 2020 in January. Facial recognition can be viewed as a core technology under the larger umbrella of AI. The framework says that human involvement must be paired with an AI framework for accountable decision-making. AI decision-making must always be explainable, transparent, and fair. It has encouraged private enterprises to join this framework and share their views.
Key concerns that must be addressed include the commitment to data always being kept safe and secure with a trusted guardian and safe-keeper, in this case the Singapore government. Technology must be rigorously tested and achieve high accuracy before it’s deployed widely to prevent misidentification and to understand its limitations. Data must only be used for specific and stringent purposes, such as for law enforcement, but not unlimited and unrestricted surveillance.
It is important for governments to be very clear about how facial recognition technology will be used. For example, Singapore is using it for its national digital identity program called MyInfo, as well as for biometric identification at borders. Each government must move carefully and maintain a balance between convenience, data privacy, and security. An understanding of the trade-offs involved is required.
Kr: Singapore is certainly very active in this area. It wants to expand the National Digital Identity (NDI) initiative beyond surveillance and security. What might that entail?
FW: We have already seen facial technology being widely used at Changi Airport’s Terminal 4, as well as at air, sea, and land border immigration. It’s also used in congested public areas like HDB public housing, the MRT, and bus interchanges. Apart from law enforcement, it’s also used to speed up immigration checks. At Singapore airport’s Terminal 4, the entire immigration and check-in process is automated, with very few human interactions required. It’s very fast and convenient for travelers.
As part of the NDI initiative, the goal is for the government to maintain a national biometric database by 2025. By doing this, it removes the need for people to remember passwords or even identity card numbers, and we will begin to see broader use cases like this in government services. One example is entry to ministry buildings or registration for e-government services, school and work attendance, high-security clearance areas or workstation access, and more.
Having been involved in the bidding of three Smart Nation projects prior to my current role, I can say confidently that the Singapore government is very concerned about rigorous and extensive testing before any technology is deployed in public. They are committed to understanding potential limitations, as well as data privacy and security implications.
Kr: Following the prevalent use of coronavirus-related surveillance, do you think that AI companies such as yours will have more leverage after this crisis?
FW: Yes, the use of AI and facial recognition will accelerate in 2020 and beyond. Businesses are now looking to cut costs, increase cash flows, and improve productivity and efficiency across operations. These are all areas where facial recognition and AI technology can really help. Crucially, even after the current coronavirus crisis, I think this will be the new way all businesses will have to operate.
Let me share a simple example from one of the industries we work with, which is banking and financial services. In Indonesia, a top digital lending platform, Danamart, used our facial comparison and “liveness detection” solution to accelerate their customer verification process, which saved them time and money, with up to 99% accuracy. This is more accurate than their human checks could ever be. Moreover, it was available 24/7, meaning it could be done any time, night or day, or during weekends. Danamart was able to redeploy staff to other critical areas of the business, saving precious time and money.
In my area of work and expertise, which is using AI in sectors such as banking, financial services, payments, retail, and e-commerce, we focus on a few key areas. The first is AI solutions which include eKYC, intelligent process automation, and chatbots. Then there is risk management, which include alternative credit scoring as well as fraud detection and prevention. Our digital lending capabilities include digital onboarding, smart decision engines, and smart collection systems.
All of these help businesses speed up their digital transformation, automate processes, and ensure fraud prevention. This allows precious resources—speed, cost, and human capital—to be redeployed across the organization for greater efficiency.