Zhu Xufeng: How to effectively combine biometric identification technology and urban governance?

A student at Northwestern Polytechnical University registers personal information in front of the terminal of the library's face recognition system. Xinhua News Agency

In recent years, digital empowerment of urban management has become the main development trend of international and domestic, and major biometric technologies such as face recognition, fingerprint recognition, and iris recognition have been widely used. What are the application scenarios of biometrics in urban governance? What are the risks? How to build and think about the governance model that effectively combines technology application and urban management? Sike invited Zhu Xufeng, professor and executive dean of the School of Public Policy and Management of Tsinghua University and deputy director of the Research Center for Science and Technology Development and Governance of Tsinghua University, to analyze.

What are the applications and risks of biometrics in urban governance?

There are four main aspects of the application scenarios of biometric technology in urban governance.

The first is the urban brain, strengthening urban traffic governance. The city brain aims to solve the traffic ills of the city's "well-developed limbs and simple mind". Through the urban integrated computing platform, the urban data resource platform and the artificial intelligence (AI) open service platform, the traffic brain identifies drivers and pedestrians, as well as traffic conditions and accidents, through face recognition and image recognition.

The Sonar Noise Acquisition System can capture and capture vehicles with sonar dense recognition technology, voiceprint comparison recognition technology, and high-definition cameras. The video and photos of the moment of violation are recorded and uploaded to the illegal horn capture management platform synchronously, and the illegal horn vehicle is accurately checked and controlled. Up to now, The City Brain has covered more than 20 cities around the world.

The second is the "Skynet Project", which serves the security field to improve public safety. "Skynet Project" refers to a video surveillance system that conducts real-time monitoring and information recording of fixed areas in order to meet the needs of urban security prevention and control and urban management.

In the fields of fugitive pursuit, personnel deployment, identity information retrieval and other fields, biometric identification technology is often used for rapid personnel comparison and confirmation, and thus becomes an important basis for security decision-making.

The data collected through the "Skynet Project" such as face recognition, fingerprint recognition video, audio, images and documents are accurately analyzed and modeled on the self-developed AI empowerment platform, forming a public security system Skynet, accurately arresting people at large, and ensuring urban public safety.

The third is population management and accurate identification of social security applications. Biometric identification technology is based on the unique biometric value of the human body for identity authentication, with unique, stable, reliable and other characteristics. The use of this technology can effectively solve the current identity authentication difficulties in social insurance management, especially the difficulties in treatment qualification certification, pension fraud and other prominent problems, to meet the current demand for real-name authentication, qualification certification and identity authentication, etc., so as to accurately identify the insured personnel.

The fourth is social governance, first of all, the identity confirmation of the invigilator, fingerprints, faces and other biometric recognition technology for the identity verification of the examiner, which can effectively eliminate the behavior of substitute examination. Secondly, it is a non-on-site law enforcement governance action, using face recognition and other technologies, pedestrian illegal behavior capture, electric vehicles taking the motor vehicle lane, reverse driving, running red lights and other serious violations of traffic order, etc., automatic capture and exposure system, which can collect evidence off-site for traffic violations.

But biometrics also have many risks, and in summary there are four major types of risk problems.

First, technical risks, that is, the problem of identification accuracy, are prone to deep forgery. Many factors affect the accuracy of biometric technology. If the fingerprint is affected by the state of the finger itself and the state of the environment, the accuracy of fingerprint recognition in different scenarios such as wet hands, oil hands or low temperature and strong light will be different; In the case of cross-channel recognition, audio noise, and multi-person scenes, the accuracy of voiceprint recognition may be reduced, and the sound changes caused by personal age and health conditions may also affect the recognition accuracy.

Deep forgery, which combines deep learning and forgery techniques, is a technology that uses deep learning algorithms and face recognition technology to create highly realistic fake images and videos. Deep forgery technology can currently use face recognition algorithms to train GAN networks and mimic facial features in face images, allowing for successful video synthesis with just one photo.

The second is data risk, and there is a security problem of excessive collection and storage. In addition to improving the accuracy of technical identification, large amounts of data as an asset itself can generate great value, so enterprises often have the motivation to collect as much data as possible and subsequent collation and analysis, resulting in the potential risk of excessive data collection.

With the development of global asset digitization and the sharp increase in the value of data, data leakage accidents are frequent, and the granularity of the leaked data is becoming more and more refined. Enterprises with massive amounts of data are facing higher data security risks.

The third is ethical risk, there is a problem of privacy violation and technological inequality. Biometrics technology is not only limited to the identification of individuals, but also can identify the user's identity, gender, age, psychological state, income information and other rich tag information. The use of this identifying information in various scenarios can lead to a series of privacy breaches.

Biometric recognition technology is the infrastructure of massive data input, and the differences in the accessibility of different social groups may affect the structure and characteristics of big data, resulting in differences in the accuracy of technology identification between different social groups, and may further lead to bias and gaps in algorithms.

The fourth is social risks, including property loss risks. Due to the inertia of the user's technical use, many users will habitually open the face recognition or fingerprint recognition function in a wide range of smart phone apps loaded with face recognition or fingerprint recognition. Because biometrics are immutable and there is a risk of forgery and imitation, once an event such as fingerprints are forged or faces are forged, the property safety of citizens will face a full range of risks.

What is Agile Governance? What are the characteristics?

Emerging technologies have some special characteristics, technology is driven by market demand, the development of technology is particularly fast, if the government policy is too flexible, the safety problems brought by industrial development may bring huge risks and even losses to society; But without resilient industrial development, innovation in emerging industries could be stifled by an excessive and outdated regulatory framework. Therefore, there is a dilemma in government regulatory policy.

We propose the concept of agile governance. Agile governance includes a set of actions or methods that require governments to be flexible, liquid, flexible, or adaptable; The second is to acknowledge that technological changes and disruptions are faster and more complex than before; Third, it not only means that the speed of governance response increases, but also needs to rethink and design policy processes.

The World Economic Forum rethinked policymaking in the Fourth Industrial Revolution by proposing the concept of agile governance in 2018. Agile governance aims to build a governance model that responds quickly and sensitively to public needs to improve organizational operational efficiency and improve user experience.

If the concept of agile governance is applied to the city, including governance objects, governance rhythms, governance methods, governance relationships, etc., it is necessary to infiltrate the agile culture into multi-level departments, form an agile organization, and form a working method and solution idea in the grass-roots public policy decision-making, public policy implementation, public problem management and public services.

The concept of agile governance was proposed to cope with the rapid development of technology. The development of technology, especially the development of biometric technology, is changing rapidly, so the uncertainty of technological development and the rapidity of technological development make it very difficult for the government to govern this technology. Governments need to be agile, able to respond quickly to the risks posed by technology, reduce the ethical and governance risks arising from new technologies, and reduce public concern by responding to technological developments with high frequency.

Agile governance recommendations for biometrics in urban management

First, adhering to the "encourage development, prevention first" technology development strategy positioning.

The government should increase the purchase of services by the public sector, support the wide application of biometrics in different areas of urban governance, and improve the efficiency and intelligence of urban transportation, public security and administrative services; Citizens should raise awareness of the security of biometric data and fully exercise the basic right to know data, so as to protect their data and privacy rights while using biometrics in the process of urban governance.

Second, actively participate in the public-private dialogue on the governance of urban biometric technology applications.

Introduce public-private dialogue, incorporate multiple entities such as providers and applications of facial recognition technologies such as enterprises and the public into the communication and cooperation system, and encourage society to widely discuss ethical boundaries; As a social force and a market force, the public can jointly identify risks and supplement policy risk research and judgment, and promote the formation of a fully responsive institutional system and policy system by the government.

Third, pilot citizens' data dividend rights in areas with conditions and needs, and leverage policy experiments on government use rights, enterprise development rights and citizens' dividend rights.

Set up areas with conditions and needs as policy pilot areas, pilot citizens' data dividend rights, and explore commercial development in the process of combining government urban management outsourcing and enterprise data model training. It is recommended that the government clearly define the boundaries of the use of public data in advance, encourage some regions to authorize the part first, and enterprises use data information authorized by citizens to train models, thereby enhancing technical accuracy and stability.

Fourth, support the government in developing contingency plans to address the social risks of biometric technology.

In extreme cases involving the risk of data leakage, data identification errors, major ethical issues, etc. in important urban service scenarios, the government may have the right to take over temporarily, and partially or even fully take over the relevant business of the enterprise.

Enhance risk awareness and collective security awareness in the face of emerging technologies, fully understand the government emergency plan based on national security and social risk control considerations, and promote the improvement of the technology and industrial risk emergency management system.

Producer: Tang Xinyi

Editor: Sun Hui

Proofreader: Zhou Jiamiao