Shanghai Jing'an District Shibei High-tech Enterprise AI Experience Center. Xinhua News Agency

The current development of artificial intelligence mainly benefits from machine learning technology represented by deep learning, which allows computers to learn and evolve independently from a large amount of data, thereby making more efficient, more accurate and cheaper predictions and decisions than humans. Because of this, artificial intelligence, as a new general technology, is believed to have a transformative impact on the economy and society. It has been elevated by countries to the height of national strategies and scientific and technological sovereignty, becoming a new focus of the ever-increasing global scientific and technological competition.

At the application level, artificial intelligence has penetrated into all walks of life. Algorithms help us filter out spam, recommend songs we may like, translate different languages ​​and words for us, and drive cars for us. Since the outbreak of the new crown pneumonia epidemic, artificial intelligence has made its mark in assisting medical diagnosis and new drug research and development, and new models such as unmanned logistics and self-driving cars have helped the development of contactless services. In short, artificial intelligence has great potential and can become a force for good, which not only brings economic growth and enhances social welfare, but also promotes sustainable development.

But at the same time, the negative impact and ethical issues of artificial intelligence and its applications are becoming increasingly prominent, calling on people to pay more attention to the ethical perspective outside of technology and its industrialization. For example, artificial intelligence model training and its application cannot be separated from the support of a large amount of data, which may lead to illegal and irregular or excessive collection and use of user data, deepening the tension between artificial intelligence and data privacy protection; facial recognition technology in some scenarios The application has also caused controversy at home and abroad about the invasion of personal privacy by this technology. Artificial intelligence technology may also be abused by criminals, such as using it to engage in cybercrime, produce, disseminate fake news, and synthesize false images that are enough to disrupt audio-visual.

With the continuous fermentation of algorithmic discrimination, the issue of fairness in artificial intelligence participating in decision-making has also attracted much attention. Studies have found that many commercial facial recognition systems have prejudice such as race and gender. Such technology is used in autonomous vehicles, which may cause dark-skinned people such as black people to be more likely to be hit by autonomous vehicles. The application of artificial intelligence in recruitment, advertising, credit, insurance, medical care, education, judicial trials, criminal sentencing, public services, etc. is also accompanied by fairness disputes. In addition, the intellectual property protection issues of artificial intelligence are also emerging. At present, artificial intelligence can independently create poetry, novels, pictures, videos, etc., and the intellectual property system will need to respond to the protection issues of artificial intelligence creations as soon as possible. Once an accident occurs in artificial intelligence applications such as autonomous vehicles and AI medical products, they will face the problem of who will take responsibility. Finally, the application of artificial intelligence may replace some manual and repetitive labor, bringing certain impact on workers' employment.

2020 is considered the first year of artificial intelligence regulation, and the regulatory paths adopted by the United States and Europe are very different. The "limited supervision" model proposed by the European Union's "Artificial Intelligence White Paper" relies more on pre-regulation and considers setting strict conditions for technology development and application; the "limited supervision" model proposed by the United States' "Artificial Intelligence Application Supervision Guide" relies more on standards , guides and other flexible means, consider setting up a safe haven for artificial intelligence applications, "regulatory sandboxes", etc. Against the backdrop of increasingly fierce global technological competition and increasingly digital economy becoming the core competitiveness of the country, considering the actual development of my country's science and technology industry, my country needs innovative governance of artificial intelligence, and rely on a variety of means such as agile supervision, ethical governance, and technical governance to jointly ensure that Artificial intelligence is being applied and developed towards goodness.

First of all, supervision not only needs to classify and classify artificial intelligence applications and be guided by problem and risk prevention, but also needs to be agile and adaptable. A core feature of artificial intelligence technology is rapid development and iteration. The formulation of mandatory laws will inevitably not keep up with the pace of technological development, so most foreign countries adopt "soft laws" such as issuing guidelines and ethical frameworks. In addition, the development and implementation of artificial intelligence applications such as autonomous vehicles and intelligent medical care still faces many legal and policy obstacles, and it is necessary to consider revising obstructive laws and regulations, and at the same time, it is promoted through "digital sandbox", "safe harbor" and "pilot" methods such as "digital sandbox", "safe harbor" and "pilot". Testing and application implementation.

Secondly, adopt ethical governance and transform ethical principles into ethical practice. At present, many technology companies at home and abroad have issued the ethical principles of artificial intelligence, and are also implementing their ethical principles through ethical review committees and ethical embeddings. These ethical governance measures in the industry have gained great consensus at home and abroad, and can better adapt to the development of AI technology.

Again, use technical means to prevent the abuse of artificial intelligence. For example, as an artificial intelligence application, deep synthesis has great positive application value in film and television production, education, medical care, entertainment and other fields, but it may also be abused by criminals to create and disseminate false images to engage in fraud and deception activities. In this regard, the industry has been actively developing and deploying content identification and traceability technologies to combat the abuse of deep synthesis. Faced with the increasing complexity and iteration speed of artificial intelligence applications, technological governance will play an increasingly important role.