Artificial intelligence is an opportunity and a challenge. Experts suggest that supervision should be strengthened from all aspects.

CCTV News:As an important part of Cyber Security Publicity Week, the 2023 Cyber Security Expo is being held in Fuzhou. The reporter noted that how to better develop the artificial intelligence industry attracted the attention of the participants. Many experts believe that although generative artificial intelligence can greatly facilitate and improve human life style, it also poses a new challenge to network security and needs to strengthen supervision from all aspects.

Experts pointed out that any technology has two sides, and there are a series of security risks hidden behind generative artificial intelligence, which pose new challenges to national security, social security, network security and data security.

In terms of security, experts pointed out that excluding human intervention, the security problem of artificial intelligence itself is also obvious.

China computer federation (CCF) Computer Security Professional Committee Member Degree:First, at the source, the data may be poisoned and polluted, so the original training data may have problems. Second, when training, there may be a big data discrimination algorithm in it, and it may not be a fair algorithm, which will lead to problems in the results. Thirdly, there will be a feedback mechanism in the final output stage, and the feedback mechanism may inject some malicious data or samples, which will also lead to the deviation of the answers obtained by this model.

So, how to deal with the hidden dangers of artificial intelligence? Experts said that this requires joint efforts from the supervision, practitioners and enterprises.

Li Xinliang, Expert of Guangdong Digital Government Network and Data Security Emergency Response Expert Group:In view of the security risks that AI may bring, I personally feel that it is necessary to speed up relevant legislation at the national level and formulate relevant industry standards as soon as possible. In addition, for AI practitioners, it is necessary to strengthen the corresponding training so that they can carry out it at the level of national laws and regulations when using AI technology. From the enterprise level, we must fully consider the compliance of collecting citizen information, and at the same time, we must strengthen the relevant actions of data protection in the transmission and storage of these collected data.