Journalism continues to increase AI implementation
An AI virtual news anchor developed by China’s tech firm Sogou delivers news at the 2020 Consumer Electronics Show in Las Vegas on Jan 8, 2020. Photo: CHINA DAILY
In the era of artificial intelligence, will computers be able to emulate the functions of the human brain? Will AI surpass human intelligence? Will robots replace human workers? These questions linger in the areas of science, technology and society. For journalism, as more AI reporters and editors engage in news production, there is a widespread concern that AI journalism might replace human journalism in the future. Or, we wonder, if not completely, to what extent can it replace human journalism?
Development of AI journalism
News activities refer to a range of human behaviors that collect, process and release information to monitor and shape the social environment. The emergence of AI journalism is a result of the development of deep learning, neural networks, intelligent algorithms and automation technology. New technologies, especially data technologies, are increasingly involved in news production.
The application of data technology is vital for the news industry in the following aspects. First, with the development of digitalization and networks, the digital traces of human beings are ubiquitous and easily accessible through large-scale distributed monitoring, constituting an important source of news content.
Second, with the rise of big data, audio-video data and complex data, data mining technology can automatically filter media content, making rapid and in-depth analysis possible.
Third, through visual presentation technology, automated writing technology, and immersive experience technology based on virtual reality and augmented reality, AI can automatically transform data into readable and visualized news narratives, allowing for innovation with the way news reports are produced.
From the perspective of news dissemination behavior, the subject changes from professional practitioners to the public, then from the public to robots, while the motive also shifts from fact presentation to social interaction, then from social interaction to value realization.
Computer-assisted journalism, digital journalism and automated journalism—the integration of computing into journalism has strengthened every step of the way. Looking back at the early 1950s, when the computer industry was budding, newsrooms had begun to use computers as a means to carry out background research.
Later, with advances in data mining technology, digital journalism gained momentum, exemplified by the use of mining algorithms to uncover hidden insights in huge data silos of structured and unstructured random data.
For example, Kira Radinsky, a data mining AI researcher, applied data analytics tools to 150 years of articles from the New York Times and other sources and discovered that in poverty-stricken areas if a year of drought is followed by a year of floods, there is a high probability of an eruption of a cholera epidemic. Her paper was shown to have predictive power because it was published several months before the severe cholera epidemic in Cuba in 2012 that claimed many lives.
Automated journalism describes the current stage in this evolution, in which intelligent algorithms are employed throughout the whole process of the journalistic endeavor, from information retrieval and the search for hidden insights through the automatic conversion of data into news stories.
In 2002, Google developed an AI editor for its Google News Service. The algorithm can crawl through thousands of news websites without human intervention and automatically select a lead story to be displayed on the home page with relevant links.
More interestingly, in 2007, a research team from the University of Tokyo created a 3D humanized robot journalist that can mingle in a crowd of people, conduct interviews, take pictures and then compose a story. The robot could explore the real world and send the information to a “news classifier” who calculated a “news score” based on the information’s “scarcity” and “relevance.” If the score was sufficiently high, the “article generator” would produce an article. Its appearance makes robot journalism more vivid and more capable of replacing human reporters.
Automated journalism was first applied in sports news, then financial news, and then quickly applied in various fields of news reports. The Associated Press publishes thousands of AI-written stories every quarter, and the New York Times has had a clear and straightforward response to the shift in journalism: “Hire more engineers.”
At the same time, drone journalism is also making its way onto the stage, with the establishment of the Professional Society of Drone Journalists in 2011.
Compared with human journalists and editors, AI journalism has great advantages in efficiency, data processing, long-hours of operation and no spatial restraints. In fact, AI agents, AI editors, and AI article generators are increasingly assuming roles that have been traditionally associated with human journalists and editors. No wonder that news organizations are recruiting more data engineers and new media operators, instead of talent who have studied journalism in traditional liberal arts schools, which leads pessimists to believe that traditional journalists will be replaced by AI altogether someday.
Limitations of AI journalism
With the large-scale application of AI technology, the problems and limitations of automated journalism begin to surface. In particular, biased and non-transparent algorithms have drawn wide criticism. People have come to realize that professional journalists must act as the guardians of intelligent algorithms in order for the algorithms to work well.
The limitations of automated journalism are mainly reflected in four aspects. One is the lack of empathy, an essential ingredient in reporting stories in newspapers. AI reporters lack the ability to comprehend complex human emotions, so it is difficult for them write news pieces that relate to readers.
Second, in-depth and investigative reporting requires reporters to delve into complex social realities and uncover the story behind the scenes, an area in which AI have difficulty proactively blending into society and engaging with different parities.
Third, AI is superior in rationally tackling ordered and complex problems, but limited in irrational thinking such as intuition and association, which are the gist of creativity. In other words, algorithms can be highly intelligent, but they cannot exceed the ceiling set by their programmers and therefore cannot be expected to produce creative stories they have never seen before.
Last, news activities not only represent the record of social reality, but also shoulder the responsibility of guiding public opinion. The latter is mainly embodied in the selection of reporting angles and news comments. However, robots have yet to master this type of news writing characterized by the close interaction of values, knowledge and experience.
Due to some inherent limitations of AI algorithms, human journalists have important advantages in competing with automated journalists. The role of journalists in the newsroom will have to adjust. As AI reporters free human reporters from repetitive, mechanical, dangerous work and all kinds of interview preparation, editors and journalists will reorient themselves toward long-form journalism, feature interviews, analysis, data-driven journalism and investigative journalism that the machines so far have not been able to generate, so that in the end human journalism will achieve unprecedented development.
Advantages of human reporters
Going forward, human journalists must learn to think out of the box and work with peer colleagues as well as AI tools.
A clever AI may excel at chess, but it is yet to be able to decode intricate philosophy. In this aspect, the advantage of human journalists is that they sharpen their thinking and come up with empowering stories that address realistic issues at their roots.
AI researchers believe that the advantage of replacing human decision-making with algorithms as much as possible is that algorithms can avoid “noise” and they can also be highly rational and rather precise. However, this view overlooks the uniqueness of human nature. Within noise, there hides surprise and humor, and an important hint of empathy. Therefore, the only way for human journalists to surpass robots is to better conform to, demonstrate and relate to human nature.
Furthermore, in today’s era, with information overload, the public, especially generation Z, have higher requirements for the form of news reports, preferring fewer texts and more emphasis on integrating video clips, new technologies like VR and AR, and new interactive applications.
Israeli communication scholar Noam Latar argued that journalism is a combination of art and science. The artistry of journalism lies in discovering new ideas for creation, looking for new perspectives for reporting, exploring new schemes for problems and opening up new modes of entertainment. The scientific nature of journalism lies in the use of analytical tools to support and process information based on the recording and storage of data on human activity. In practice, the latter can be taken over by AI, whereas artistic work must rely on human reporters.
In the wave of technology-driven development, human beings should make clear the boundaries and rules of technology application. The more advanced the technology, the more dependent human beings will be. However, people will be alienated, marginalized and atomized in the creation of science and technology if they only ask “whether there is scientific basis” but not “whether there is humanistic care.” AI without a sense of humanity is at best machine intelligence, not human intelligence, and at worst it might reduce the overall level of human intelligence.
Hu Yu is a professor from the School of Journalism and Communication at Tsinghua University.
edited by YANG XUE