Bid farewell to improper Manipulation of Internet algorithms recommended "wear tight bands"
Influence network public opinion, big data "kill", induce addiction... With the development of technology, the dark side of Internet algorithms began to be exposed. These issues are now subject to heavy regulation. On January 4, the Cyberspace Administration of China (CAC) and other government departments jointly revised the Regulations on the Management of Algorithm Recommendation for Internet Information Services (hereinafter referred to as the Regulations), which stipulates that algorithm recommendation shall not be used to influence online public opinion or to kill people.
"Kill well" or become a past tense
In terms of content, compared with the draft for soliciting opinions, the official release of the Provisions has 35 articles, including 5 more articles, involving the standardization of information services and the protection of users' rights and interests. On March 1 this year, the regulations will officially come into effect. This is China's first targeted algorithm recommendation rules and regulations in the field of Internet information services.
What is algorithmic recommendation? The "Provisions" first defines it as providing users with information by using algorithm technologies such as generative synthesis class, personalized push class, sorting and selection class, retrieval and filtering class and scheduling decision class.
Among them, the algorithm recommendation information service provider plays an important role. According to the Regulations, platforms that provide users with information services by using technologies such as generation and synthesis, personalized push, sorting and selection, retrieval and filtering, and scheduling and decision-making are all classified as algorithmic recommendation information service providers.
Algorithms have always been synonymous with efficiency. The algorithm draws different portraits according to the user's age, occupation, habits of interest, consumption level and other characteristics, so as to accurately screen the massive information, and directly push the information that the user wants to see and the service that accepts the demand to the user. Travel, online shopping, entertainment, these infiltrated into the daily life of algorithm recommendation applications make users feel convenient.
But somewhere along the way, the evolution of algorithms began to sour. With careful recommendation followed, is the price monopoly, big data "kill familiar" and other users most often joked about the problem.
"The same starting point on taxi-hailing apps shows different prices for different devices." "Products that you don't care about at all, you click on by mistake, and then they appear frequently on the recommendation screen, and even appear in 'thoughtful' cross-platform ads." The experience that the netizen makes fun of seems to be multifarious, but is similar with minor differences.
"Almost all web companies are run by algorithms. The difference is just how advanced the algorithms are." Wang Yuan, a partner at Beijing Gaoqin Law Firm, told Beijing Business Daily that, simply put, algorithm-based recommendation can generate content through a series of professional data calculations, which can influence people's thinking and even behavior.
In view of the above problems, the Regulations require that, according to the preferences of consumers, trading habits and other characteristics, the use of algorithms in trading prices and other transaction conditions to implement unreasonable differential treatment. This also means that using big data to "kill" may become a thing of the past.
It shall not influence online public opinion
Big data "kill" is visible manipulation, and algorithm recommendation out of control brings far more problems.
"Algorithms shall not be used to influence online public opinion, evade supervision and management, and conduct monopoly and unfair competition." In a series of details specified in the regulation, "Online public opinion shall not be influenced" is particularly prominent.
The list under the control of the algorithm is a good example of public opinion influenced by the algorithm. Beijing Business Daily reporter learned that at present in many information information and short video platforms, drainage, limiting the flow, manipulation of the list has become a routine operation of content presentation.
In this regard, the regulations point out that providers of algorithm-based recommendation services shall not interfere with information presentation, influence online public opinion or evade supervision and management by using algorithms to block information, over-recommend information, manipulate lists or search results ranking, control hot search or selection.
"It is necessary to regulate algorithmic recommendation information service providers, to clarify which units are within the scope of national regulations and which are not, so as to prevent people with ulterior motives from using public resources to achieve private goals." "We media practitioner Ms. Yao admitted.
Besides inadvertently by ranking of induced form of thinking, by big data "kill" influence behavior, the Internet vulnerable populations, such as minors, the elderly group, and the influence of the algorithm is more directly, such as short video platform using the algorithm continuously push addictive video, while criminals loophole induction of old people fall into the catch.
In this regard, the "Regulations" also made clear provisions, should not use algorithm recommendation services to induce minors to indulge in the Network, should facilitate the safe use of algorithm recommendation services for the elderly. Order allocation, remuneration composition and payment, working hours, reward and punishment algorithms should be established and improved on the platform.
Extending to all groups, the Regulations specify relevant requirements on the protection of users' rights and interests, including protecting users' right to know and choose algorithms, informing users of their provision of algorithm recommendation services, providing users with options that are not specific to their personal characteristics, or conveniently shutting down algorithm recommendation services.
Landing still requires multiple efforts
The implementation of the Regulations is on the way, but the concrete implementation of the regulations still requires the coordination of various parties to make the regulation more fair and transparent, permeating the whole process.
Wang yuan told Beijing Business Daily, "Algorithm recommendation is a new technology, and relevant departments have issued strict laws and regulations for this technology, but there is no landing grip at present. In terms of information delivery service platforms, there is still no standard guidance on how to classify and grade them. Industry practice often takes the lead, and relatively speaking, regulation shows a certain lag, but after the "savage growth", algorithm recommendation is calling for regulation.
According to the Regulations, algorithmic recommendation platforms providing Internet news and information services should obtain service licenses in accordance with the law, which may provide an antidote for the regulation of algorithmic recommendation.
In addition, "filing procedures" also become a powerful shackles. The regulations specify that providers of algorithm-based recommendation services with public opinion attributes or social mobilization capabilities should go through filing procedures within 10 working days from the date of providing services, and carry out safety assessments in accordance with relevant state regulations. At the same time, it shall keep network logs in accordance with the law, cooperate with the cyberspace administration and relevant departments of telecommunications, public security and market supervision to carry out safety assessment, supervision and inspection, and provide necessary technical and data support and assistance.
"The governance of algorithm recommendation service requires the joint participation of the government, enterprises, society, netizens and other parties to promote the fairness, fairness, standardization and transparency of algorithm recommendation service, promote the upward improvement of algorithm recommendation service, and create a clearer cyberspace." A person in charge of the CAC said that at present, the national and local CAOs are taking the lead in coordinating the governance of algorithm recommendation services and related supervision and management work across the country.
National laws also provide a strong backstop. The reporter learned that the Personal Information Protection Law, which will take effect on November 1, 2021, systematically regulates the excessive collection of personal information, the "killing" of big data and the handling of sensitive personal information, which are of public concern.
"In addition to administrative supervision, procuratorates and consumer associations can file lawsuits for infringement on behalf of individuals. In general, power is increasingly centralized, with each industry performing its own duties under the overall control of the CAC." Hanah said.
Reporter of Beijing Business Daily & NBSP; Yang Yuehan & have spent Intern journalist & NBSP; WeiLu