Towards Man-machine collaboration: Journalistic ethics in the era of algorithmic journalism

With the rapid development of artificial intelligence today, the news pattern is also being reconstructed. Algorithmic journalism is gaining momentum. In algorithmic journalism, the problem of journalistic ethics anomia caused by algorithmic bias and information cocoon not only makes us call for the return of journalistic professionalism, but also makes us find a new path in the tide of algorithmic journalism to actively build responsible news production. To adhere to the unity of instrumental rationality and value rationality and to establish the mode of man-machine collaboration is a path for us to rebuild news ethics.

With the rapid development of science and technology such as artificial intelligence, the Internet and big data, the field of news is also undergoing great changes. The traditional news model from the print era is shifting to the more interactive and experiential intelligent media model in the "Internet +" era. As a technology, artificial intelligence algorithm plays an increasingly important role in the process of news production and push. The pursuit of algorithmic journalism is often accompanied by controversy, especially at the level of journalistic ethics. In recent years, we can often see different opinions from all sides in the algorithmic news discussion. The development trend of algorithmic news is inevitable. How news media can make use of the dividend brought by this technological revolution to give news production new value and significance is the concern of all news industry stakeholders. Therefore, the author hopes to explore how to make algorithmic news achieve more benign development by analyzing the challenges faced by news ethics in the new situation.

First, the rise of algorithmic journalism

(A) What is algorithmic news?

In recent years, there have been more and more discussions on the concept of algorithmic news, which has gradually become a professional term. Expressions about it include robot news, automated news, data-driven news, computational news, and algorithmic news. Some argue that the concept of "algorithmic journalism" is more rigorous. Algorithmic news is a process, method or system that uses intelligent algorithm tools to automatically produce news and realize commercial operation, including the automation of information collection, storage, writing, editing, display, data analysis and marketing. This expression more accurately reveals the essential characteristics and basic laws of the new generation of news production. [1] It specifically includes news compilation, algorithm recommendation, aggregation and distribution of news platforms and other processes, which is an automatic news production mode different from traditional media. Its core content is a set of algorithmic mechanisms applicable to the news production process, that is, the proportion of technology in news production has been strengthened unprecedentedly.

Philip Hammond, co-founder and CTO of automated writing pioneer Narrative Science, predicted in 2011 that 90 percent of press releases would be written by computer algorithms within 15 years. Algorithmic news has developed rapidly in recent years, showing the great impetus of big data and artificial intelligence technology in the news field, which also represents the trend of future news production.

Compared with the past, the difference of algorithmic news is that the changes it brings to news production are essential, and its core content is mainly reflected in two aspects:

First, the author of traditional news is human, while the author of algorithmic news is machine. In the age of traditional media, news gathering, editing and pushing rely on reporters and editors. After entering the era of algorithmic news, the production process of news includes data capture, information processing and analysis, news compilation, news push and distribution. This process can be done through the automation of algorithms.

Second, news feed adopts algorithm recommendation. In the past, paper media and portal websites mainly distributed and pushed news by human editors. In the era of algorithmic news, there are more news aggregation platforms. They classify the user group through the analysis of big data, and use the recommendation algorithm to classify and push news, so as to provide personalized news services for the reader market. This has undoubtedly become a major force in the current media production model.

(2) Practice and controversy of algorithmic news

The technical application of algorithmic news has been developing in recent years, from the initial robot writing attempt to the widespread market application of news recommendation algorithms, algorithmic news has brought more challenges to traditional media. In the European and American countries where algorithmic news was first applied, the debate on the advantages and disadvantages of algorithmic news becomes more and more intense with its great expansion in the field of news production, and even brings more contradictions and conflicts.

One is the promotion and application of robot writing. In 2006, Thomson Financial began using computer programs to write financial news. This caused concern at the time, and questions followed. In the years that followed, robot writing grew rapidly, and several Automated writing companies began to emerge, such as Narrative Science, Automated Insights, Yseop, and others. In the field of media, the Associated Press, Washington Post, Los Angeles Times and other media compete to use robot writing, Blossom, Heliograf, Quakebot, Wordsmith and other news writing software have been widely used. Its fast, accurate and efficient operation mode has indeed brought great convenience and benefits to news production. The pace of robot writing in China has also accelerated in recent years. In 2015, Tencent launched its first news report written by Dream writer, a robot writer. Xinhua News Agency's "quick pen Xiaoxin" and "Zhang Xiaoming" of "Today's headlines" have been on the job. Bot writing is the earliest and most dominant form of algorithmic journalism. It presented a powerful advantage from the moment it appeared. The first is its efficiency. From taking the lead in financial and sports news, to expanding into more areas, from the initial programmatic writing, to trying to provide personalized news. The second is the automation of news production. It makes it easy to capture and compile data accurately and quickly, freeing journalists from repetitive trivial labor.

However, robot news writing was questioned from all sides almost as soon as it was put into news production. First, because the technology is driven by software and algorithms, it is considered unable to show the deep meaning behind the data, that is, it lacks depth and personality. Second, it is also criticized for its lack of verbal flexibility and emotional expression, unable to reflect the unity of ideas and news information, and lack of direct communication with interviewees, especially emotional communication. For this reason, some people call it news without temperature. Third, robot writing can now develop personalized writing templates according to customer needs. However, the process of entering information on this template also includes some preliminary filtering of information. The news that the user sees may only be a measurement of the event. The practice of algorithmic journalism proves that the development of this emerging model may face more trial and error.

The second is the news feed under the design of recommendation algorithm. At present, platform media has become the main place of news feed. In order to seize the market, Facebook, Twitter and other media platforms began to implement personalized news feeds with the help of recommendation algorithms. Machine algorithms replace human editors. Through data capture and analysis, users' interests and hobbies are measured, and then the analysis and calculation results are applied to the news feed to push the news information they are interested in for different target groups. These platform media often dominate the aggregation and distribution of news products and dominate the flow of news transmission.

Recommendation algorithm news is mainly divided into: popular recommendation, according to the interests of users to recommend. Through the principle of collaborative filtering, algorithmic news recommendation is indeed more efficient and accurate. However, there are also worrying problems. Users often selectively receive one type of information, while others are blocked. The resulting "information cocoon" effect is often not conducive to the audience's comprehensive and objective understanding of social public events. Users are immersed in the information circle under the algorithm recommendation, forming a closed environment, and the barriers between different social groups will be strengthened. In addition, the replacement of manual editing by machine algorithms will also lead to misdirection and low-level spam information spread through the platform, resulting in adverse social impacts.

The third is the competition between traditional media and new media. According to a 2016 Pew Research report, most Americans prefer to read news on their phones, and more than two-thirds of Facebook users use the service primarily for news. In 2016, 66 percent of Facebook users read news or news headlines on the social network, up from 47 percent at the end of 2013. The Center's 2018 report noted that about two-thirds (66 percent) of people believe that news posted by bot accounts has a great deal of negative impact on Americans' knowledge of current events, and almost no one believes it has a positive impact. The more news the public thinks bots produce in the news environment, the more difficult it will be for people to understand the truth. In 2019, the Pew Center's report also showed that Facekook, Twitter and other media platforms have caused more public concern in terms of political and racial bias.

In 2017, the US News Media Alliance (News Media Alliance) issued a statement that Facebook and Google's data algorithms one-way determine the traffic of news; In addition, the use of many media on the network news content, and earn high network advertising revenue. This duopoly prevents the news media from providing the best quality news, and even allows users to preferentially receive inferior fake news and absorb misinformation through algorithms.

When the algorithmic "domination" of news production has exposed more worrying problems, the professionalism of traditional media people to news has been mentioned again. Sharb Farjami, CEO of Storyful, the most reliable source investigation and fact-checking organization, pointed out at the 2017 Tencent Media + Summit that human editors are still the core of fact checking, and the authenticity of news should be protected by both technology and human checking.

Indeed, algorithmic news brings new problems, such as the lack of depth and thought in news reports, the difficulty of supervision, and the "crowd classification" of users in news feeds. To improve the quality of news products, it is obviously not enough to rely on technology and algorithms alone, and it is also necessary to strengthen the professional ethics and responsibility of journalists. In this regard, journalistic ethics is indeed in an awkward position.

2. Challenges brought by algorithms to news ethics

Because algorithmic news combines new technology, artificial intelligence as the carrier, with the advantages of algorithms, so that news production and dissemination on the "highway". In this process of continuous strengthening, the subjectivity of people in news production is no longer as unbreakable as in the past. In recent years, the idea that artificial intelligence will replace human beings and writing robots will replace journalists has often caused concern and discussion in the industry. However, as mentioned earlier, in the Western countries that are at the forefront of algorithmic journalism, the audience's evaluation of algorithmic journalism is also mixed. The discussion and questioning about it mainly focus on the professional norms and ethics of journalism. Therefore, it is a very important topic to investigate the ethical problems in algorithmic journalism. In the era of algorithms, journalism ethics needs to face the following issues:

(1) Challenges to journalistic professionalism

The rapid development of algorithmic news has brought an impact on traditional news media. The first to be hit is the authority of traditional media. After long-term development and accumulation, modern journalism itself has formed a professional standard system, which is journalistic professionalism. The core of traditional journalism professionalism is to require journalists to serve the public interest of the society and provide true, comprehensive, objective and fair news reports for the society and the public. Lu Ye and Pan Zhongdang have made a special generalization and discussion on journalistic professionalism, which has gained more recognition. The expression of journalistic professionalism is as follows: "Journalism is an occupation, and when it is called a profession, we refer specifically to the specific professional skills, norms of behavior and standards of evaluation that must be obtained through professional training and agreed by journalists." The concept of 'professionalism' goes far beyond these occupational characteristics. It also includes a set of beliefs that define the social function of the media, a set of professional ethics that govern journalism, a spirit of obedience to a higher authority beyond political and economic power, and a conscious attitude of public service.

Journalistic professionalism puts forward the universal principle of journalistic ethics, which requires news media to be responsible for the objectivity and authenticity of news reports, play a positive guiding role in social public interests, and have positive social significance. Under the guidance of the standard of journalistic professionalism, traditional news media have also formed their own authority. However, since the emergence of algorithmic news, the dominant position of journalists has begun to shake. In order to pursue the maximization of interests, Internet media make full use of artificial intelligence algorithms to capture information, produce news automatically, and then push news through algorithm recommendation. The role of news media is constantly dispersed and weakened, and non-professionals can also compile news information with the help of algorithms. Even in the process of news feed, the position of the gatekeeper of traditional media is replaced by the algorithm, which makes classified recommendations according to the preferences of the audience, which leads to the weakening of the power of more traditional news media and the collapse of the authority of news media.

(2) Information cocoon

In algorithmic journalism, people have access to even greater amounts of data. But through big data analysis and collaborative filtering principles, news products are categorized and sent to different target groups. At the beginning, this way of news push may be favored by many platform media due to its precise delivery, which is believed to greatly improve the efficiency of selective reading by the audience and provide customers with more convenient and rapid information consumption. However, this also creates a further problem, which is the information cocoon effect.

Cass Sunstein, a law scholar at Harvard University, put forward the concept of "information cocoon" in his book Information Utopia: How People Produce Knowledge. In the communication of information, the public's own information needs are not comprehensive, because the public only pays attention to what they choose and the communication field that makes them happy, over time, they will shackle themselves in the "cocoon" like a cocoon. On the one hand, the information cocoon has caused a situation of "people are divided by groups", which has hindered people's comprehensive cognition of society, hindered the circulation of public information, and also hindered the promotion of public interest topics. For more people, it is like falling into a well of careful classification and selection, and people become "frogs in the bottom of the well". On the other hand, information cocoons may create more biases and misunderstandings. When users choose information, they filter out some information and stick to their own circles. The concepts of "filter bubble" and "echo chamber" effect are also descriptions of the limitations of information received by users under the algorithm mechanism from different perspectives.

(3) Algorithm bias

Algorithm bias refers to the bias of the program designer himself, which is brought into the algorithm program design, so that the algorithm appears in the application of a certain bias. The problems caused by algorithmic bias are often hidden at the beginning, but differentiating the audience, especially when it comes to race, religion, gender, and age, may cause unfair and asymmetrical news dissemination, thus benefiting some groups and damaging the interests of others. For example, Facebook's "biasgate" incident is a classic example of algorithmic bias: Facebook has been accused of manipulating its trending topics list and politically biased news screening. The incident caused an uproar.