The Provisions on the Administration of Internet Information Service Algorithm Recommendation jointly issued by the Cyberspace Administration of China and four other departments came into effect on March 1. For ordinary users, algorithms cannot be touched or understood, but they are widely affecting our access to Internet information services through various platforms.
Algorithms are everywhere Abuse problems constantly
Algorithm recommendation technology captures users' daily usage data, analyzes people's behaviors, habits and preferences, and then accurately provides various services such as information, entertainment, and consumption. While providing convenience, in recent years, unreasonable applications such as "big data killing", traffic fraud, and inducing addiction have also brought trouble to our lives. According to the "China Great Security Perception Report (2021)" jointly released by the Internet Development Research Center of Peking University and Internet companies, 70% of respondents feel that algorithms can obtain their own preferences and interests to "calculate" themselves, and nearly 50% of respondents said that they want to escape the network and stay away from mobile phones under the shackles of algorithms.
Zuo Xiaodong, vice president of the China Academy of Information Security: The push of these services is customized by network service providers, so customization is easy to abuse, light damage to our personal rights and interests, serious will endanger national security, so algorithm recommendation must be regulated.
The Provisions make it clear that the application of algorithm recommendation technology refers to the use of algorithmic techniques such as generative synthesis, personalized push, sorting selection, retrieval and filtering, and scheduling and decision-making to provide information to users. Almost all kinds of Internet companies that provide algorithm recommendation services are within the scope of supervision, such as various short video platforms, e-commerce platforms, social platforms and catering takeaway platforms.
In order to break the "black box" that the algorithm cannot touch or understand, the Provisions clearly require that the user's right to know the algorithm and the right to choose the algorithm should be guaranteed, and the user should be provided with options that do not target their personal characteristics, or the option to conveniently turn off the algorithm recommendation service.
The Provisions require that algorithm recommendation service providers should adhere to the mainstream value orientation, actively disseminate positive energy, establish and improve mechanisms for manual intervention and user self-selection, and must not use algorithms to influence network public opinion, evade supervision and management, and monopolize and unfair competition.
No differential treatment can be implemented, and the sword refers to the "killing" of big data.
In recent years, some enterprises have used personal information to engage in big data "killing", differential pricing of different groups, and implementation of "price discrimination", which makes consumers very disgusted. In this regard, how will the "Provisions" be cracked?
Ms. Hu of Shaoxing, Zhejiang Province, ordered a hotel room through a business travel platform, and when she checked out, she found that the room rate paid through the platform was nearly double the actual room rate of the hotel, and she was a senior member of the business travel platform, and she should have enjoyed a 15% discount. Ms. Hu suspected that the business travel platform had "big data killing" behavior through her previous consumption behavior, and sued the business travel platform in court. In July last year, the court granted the plaintiff's request for a refund of one to three in the first instance. This case has been called by many media as the first case of "big data killing".
The same delivery time, place, order, takeaway platform, but members pay more than non-members; at the same time, the same type of car to the same destination, a taxi platform has been found by users to charge more regular customers, similar consumer complaints have occurred. In this regard, the Provisions propose that where algorithms recommend service providers to sell goods or provide services to consumers, they shall protect consumers' right to fair trade, and must not use algorithms to carry out unreasonable differential treatment and other illegal acts on transaction conditions such as transaction prices based on consumers' preferences and trading habits.
Zuo Xiaodong, vice president of the China Academy of Information Security: Some platforms will say that this algorithm is my intellectual property and my trade secret, and I cannot disclose it. None of these arguments hold true, and the platform's algorithms must ensure the necessary transparency.
Experts stressed that the "Provisions" also clearly protect the rights and interests of users such as the right to choose and delete. Consumers are prevented from being "calculated" by the algorithm by providing users with the option not to target their personal characteristics, or by providing users with the option to turn off algorithmic recommendation services conveniently, and by providing the ability to select or delete user tags for their personal characteristics.
Protect the legitimate rights and interests of workers to avoid being "trapped" by algorithms
In addition, in the "Provisions on the Administration of Algorithm Recommendation of Internet Information Service", it is required that algorithm recommendation service providers providing work scheduling services to workers shall protect the legitimate rights and interests of workers such as labor remuneration, rest and leave, etc.
In September 2020, an article titled "Takeaway Riders, Trapped in the System" attracted much attention. The article describes that under the algorithm and data drive of the takeaway platform system, the delivery time of the takeaway rider is constantly compressed, and the rider, driven by the powerful system, in order to avoid bad reviews and maintain income, has to choose retrograde, running red lights and other practices in reality. Many takeaway riders said that in dozens of high-rise buildings, elevators are ten minutes or even longer, but such objective situations are not taken into account by the "algorithm system".
Liang Zheng, vice president of the Institute of International Governance of Artificial Intelligence of Tsinghua University: Since the machine is still in the development stage, for some labor work scenarios, the decision power is handed over to the machine, and there is a great potential risk of fairness and reliability.
Internet platform optimization algorithm mainly through the big data analysis algorithm model to calculate the best strategy for business operations, at the same time, with the algorithm with a set of strict performance appraisal mechanism, once violated some of the algorithm rules, it will lose the "advantage" in the assessment, and may bring some vicious response behavior. In this regard, the "Provisions" specifically proposes that algorithm recommendation service providers that provide work scheduling services to workers shall protect the legitimate rights and interests of workers such as obtaining labor remuneration, rest and leave, etc., and establish and improve relevant algorithms such as platform order distribution, remuneration composition and payment, working hours, rewards and punishments.
Ensure that the algorithm is open and transparent, and strictly prevent the brush measurement control evaluation from affecting public opinion
Hiring network "water army" to implement traffic fraud, encouraging "rice circle" fans to tear each other apart, brushing amount control and evaluation, and some algorithm recommendation service providers to intervene in hot searches, the phenomenon of affecting network public opinion is criticized. The "Regulations on the Administration of Recommendation of Internet Information Service Algorithms", which will be implemented tomorrow, will put a "tight curse" on such "black hands" that affect online public opinion.
In the entertainment industry, some traffic stars can often get hundreds of millions or even billions of commercial returns, star business platforms do everything possible to produce topics and hype, any behavior that can trigger fans' attention and discussion, and even the network scolding war between fans, is the regular planning content of the business team. The brand side is also well aware of this, and even uses means to deliberately create contradictions between different fans, setting off a scolding war to attract attention and pull traffic.
The Provisions on the Administration of Recommendation of Internet Information Service Algorithms, which came into effect on March 1, require that algorithm recommendation services should follow the principle of openness and transparency, and encourage algorithm recommendation service providers to comprehensively use strategies such as content deduplication and fragmentation intervention to optimize the transparency and interpretability of rules. Establish and complete management systems such as algorithm mechanism mechanism review, scientific and technological ethics review, user registration, and information release review, and must not use algorithms to manipulate lists, control hot searches, and other such intervention information presentation.
Zuo Xiaodong, vice president of the China Academy of Information Security: The content we see may deviate more and more from justice, deviate from neutrality, be artificially or have algorithms behind it, then necessary intervention is needed.
Zhang Xin, Executive Director of the Digital Economy and Legal Innovation Research Center of the University of International Business and Economics: In principle, tell us that these designers cannot develop such algorithms, what if such corresponding behaviors are carried out? This time, the management regulations are very clear to bear the corresponding responsibilities.
Prevent inducing addiction to consumption and protect the interests of special groups
How to prevent users from being addicted and over-consuming, and protect the rights and interests of special groups such as minors? In response to these demands, the "Provisions" that will be implemented tomorrow have also made refined provisions. Many experts said that establishing and improving a scientific regulatory system and forming a long-term mechanism for algorithmic governance will be the core essence of ensuring the implementation of the Provisions.
The Provisions propose that algorithm recommendation service providers shall periodically review, evaluate, and verify algorithm mechanism mechanisms, models, data, and application results, and must not set up algorithm models that violate laws and regulations or violate ethics and morality, such as inducing users to indulge or consume excessively.
For minors, the Provisions propose that algorithm recommendation service providers must not push to minors information that might cause minors to imitate unsafe behavior, violate social morality, induce minors to have bad habits, and other information that might affect minors' physical and mental health, and must not use algorithm recommendation services to induce minors to become addicted to the Internet. The Provisions also require that where algorithms recommend service providers to provide services to the elderly, they shall protect the rights and interests enjoyed by the elderly in accordance with law, fully consider the needs of the elderly such as travel, medical treatment, consumption, and handling affairs, and provide intelligent services for the elderly in accordance with relevant national provisions.
How to implement supervision? The "Provisions" make it clear that prior supervision shall establish systems such as algorithm filing and algorithm evaluation; distinguish supervision and establish the idea of hierarchical classification; and system supervision to establish a situation of pluralistic co-governance. Experts believe that the Provisions focus on improving the governance system and realizing the whole process and the whole chain of supervision before and after the event. If the algorithm recommendation service provider violates the relevant regulations, it will be punished with warnings, notification of criticism, fines, etc., and if necessary, public security administrative penalties and criminal liability will be pursued.
Zuo Xiaodong, vice president of the China Academy of Information Security: Emphasizing the responsibility of the main body is very important, because the platform is the user recommended by the algorithm, so only by compacting the responsibility of the platform, then the platform can consciously implement the requirements of laws and regulations.
Source: CCTV