Abstract
Keywords
Introduction
It is widely known that health-related information plays a crucial role in shaping public perception and judgment of health risks (Wang & Xu, 2020). During major public health emergencies, such as the COVID-19 pandemic, it is imperative to provide the public with accurate, reliable, and timely information. This is crucial for helping individuals understand and cope with social changes, mitigating panic, and facilitating the search for social support. For news agencies, the pandemic presents a unique opportunity to harness the potential of automated news generation through structured data.
This paper aims to explore the influence of various writers and news formats on audience attitudes toward pandemic news. Additionally, it examines how technological advancements can be utilized in news generation to achieve comprehensive, scientifically accurate, and effective dissemination of pandemic-related information.
The COVID-19 pandemic emerged as a major public health crisis in recent years. Initially detected towards the end of 2019, the virus rapidly spread across the world, instilling uncertainty and urgency due to its high infection rate. Consequently, individuals sought solace in pandemic-related news, recognizing the societal and practical importance of media during unprecented times (Li et al., 2021a, 2021b).
News agencies faced a significant challenge in effectively reporting the pandemic. The sheer volume of data related to the pandemic, combined with the need for real-time updates, rendered traditional news production methods insufficient. However, advancements in technology have presented opportunities for news agencies to leverage artificial intelligence (AI) and big data to help produce timely, informative, and accessible content.
The integration of technology in news production is not new. However, the recent pandemic accelerated the use of automated news generation. Utilizing structured data, machines can generate news reports by collecting, selecting, and standardizing information. Subsequently, algorithm models analyze and process the data based on media requirements. Finally, appropriate templates are selected, and the script is produced. In December 2018, Xinhua News Agency launched the “Media Brain-MAGIC Short Video Intelligent Production Platform,” exemplifying this technological advantage. The platform integrates various AI capabilities, such as natural language processing, computer vision, and semantic understanding of audio to enable data visualization, data-based video production, and video automation from data collection to video uploading.
Furthermore, in response to the large demand for real-time data visualization in reporting the pandemic, Xinhua Zhiyun launched a news robot to examine the stories behind the data. This study examines whether different writers and news formats influence audience attitudes toward pandemic news. Using the Cognitive-Affective-Conative (CAC) model as the framework and a 2 × 2 factorial online experiment method, the results indicate that the subject matter and news formats significantly influence audience attitudes.
Audience attitudes toward machine-generated video news is significantly more positive than news written by human journalists. This finding demonstrates that machine-generated news may break down digital anxiety and cognitive barriers, providing positive driving forces to maintain the social stability during major public health events.
The pandemic revealed the importance of media during public health emergencies. News agencies must produce timely, accurate, and accessible information to help individuals understand and cope with social changes related to the pandemic. Automated news generation presents opportunities for news agencies to leverage AI and big data to produce effective content.
This paper’s findings indicate that machine-generated news positively impacts audience attitudes toward pandemic news. The findings highlight the potential application of automated news generation in overcoming digital anxiety and cognitive barriers during major public health events. This can aid in maintaining social stability.
Literature review
Research on audience attitudes toward machine-generated news
With the increasing use of artificial intelligence in news production, the impact of different writers and news formats on audience attitudes is a critical research area. Previous studies have focused on individuals’ characteristics (e.g., education level, media literacy, and usage hours) and media characteristics (e.g., framework and stereotypes). Further investigation is warranted to explore the impact of machine-generated news on audience perceptions. This study examines findings from previous research and investigates the influence of different writers and news formats on audience attitudes toward news.
Previous studies have shown that key characteristics can impact individuals’ attitudes toward news (Hung, 2019). Additionally, it is important to note that the identity of the subject, whether they are a journalist or a news consumer, can influence the audience’s perception of news credibility (Van der Kaa & Krahmer, 2014, Zheng et al., 2018).
Research has also shown that media characteristics can impact audience attitudes toward news. Media attributes, such as visual appeal and interactivity, can influence audience perceptions and rating of news (Sundar, 2000). Moreover, stereotypes of artificial intelligence and human journalists and expectation-confirmation theory can impact individuals’ perceptions of news written by different writers (robots vs. human journalists) (Jung et al., 2017). Previous experience with robots in the media can also influence individuals’ acceptance of new technologies (Graefe et al., 2018; Waddell, 2018).
Research has explored the impact of news content factors on audience attitudes. However, audience members assign significantly different ratings to various fields and sources (Van der Kaa & Krahmer, 2014; Zheng & Yang, 2019). With the advent of natural language generation technology, machine-generated news has gradually emerged in political, social, and other news genres. Therefore, audiences may perceive news differently (Jiang & Shi, 2019).
In conclusion, attitudes toward news are influenced by a range of factors, including their basic characteristics, media characteristics, and content. With increased adoption of AI in news production, it is essential to investigate the impact of writers and news formats on individuals’ perception of news. This paper highlights the need for further research to explore the impact of machine-generated news on audience attitudes toward news.
Health information dissemination during COVID-19
The pandemic has impacted individuals and societies globally, and the Internet has played a vital role in connecting people virtually. The pandemic’s unknown nature and high transmissibility led to widespread fear and irrationality. The data released online affected people’s moods and mentalities. As a result, pandemic reporting must meet higher standards, and visualization has become an essential tool for media coverage. Different media produces various characteristics, and individuals exposed to various media exhibit differing pandemic prevention reactions. Thus, it is vital to investigate the impact of information characteristics on audience attitudes toward pandemic news. This paper explores whether journalists (robots vs. human) and news formats (video vs. non-video) affect perceptions of pandemic news.
COVID-19’s complexities and uncertainties underscored the importance of accurate and reliable data visualization in pandemic reporting. Data presentation plays a critical role in influencing individuals’ perceptions and attitudes toward the pandemic (Wang & Shi, 2021). Various media channels have different characteristics that impact individuals' pandemic prevention behaviors.
The influence of media coverage on risk perception is essential, particularly during a public health emergency like COVID-19. While previous studies have focused on attitudes toward news, few studies have explored the impact of information characteristics, especially health information transmission with various attributes during public health events (Liu & Xie, 2020).
The pandemic revealed the importance of data visualization and accurate pandemic reporting through various media channels. Data visualization and information characteristics greatly affects individuals’ perceptions of the pandemic. The influence of journalists and formats on audience attitudes toward pandemic news is critical, reflecting the salience of media technology and information formats for public perception. This study’s findings can inform future research and contains significant implications for pandemic reporting and communication during public health emergencies.
Cognitive-affective-conative model
The CAC model is widely used, it describes the formation of attitudes, early research illustrated that correspondence between measures is an important moderator of attitude-behavior consistency, with larger correlations between attitudes (affective or cognitive) and behavior when both measures refer to the same action, target, context, and time (Fishbein & Ajzen, 1977). Attitude has been a fundamental conceptual concern of social psychology. Previous research provided a comprehensive review of a wide range of the theoretical and empirical literature devoted to attitudinal phenomena, and demonstrated that this literature can be incorporated within a unified and systematic explanatory structure, that is where the CAC model comes from. In other words, attitude can be understood as the overall evaluation individuals maintain toward a specific object, based on their cognition, affection, and conation.
Cognition refers to a person’s knowledge, views, beliefs, and ideas about a perceived object, which can be viewed as a person’s perception of an information system. This cognitive component of attitude is typically formed based on an objective assessment and understanding of the perceived object’s key characteristics, and the perception consists of perceived ease of use and perceived usefulness (Lin, 2014). Perceived ease of use refers to “the degree to which a person believes that using a particular system would be free from effort,” and perceived usefulness refers to “the degree to which a person believes that using a particular system would enhance his or her job performance” (Davis, 1989).
Affection is an emotion that forms based on cognition, it involves a person’s perception and rating of an object, person, problem, or event. This emotional component of attitude can lead to a positive or negative evaluation of the perceived object. Accordingly, affective response can be treated as a person’s emotional response to an information system (Li, 2013), which may be positive or negative feelings that lead to favorable or unfavorable evaluation of the system (Kim et al., 2013).
Conation is a comprehensive behavioral tendency that forms on the basis of cognition and affection, which refers to a person’s behavioral intentions and his/her actions with respect to or in the presence of a given object (Fishbein & Ajzen, 1977). It reflects a propensity to engage in certain actions or behaviors, and it can be deemed as a person’s behavioral intention of using an information system (Kim et al., 2013). Behavioral intention is defined as “a person’s subjective probability that he will perform some behavior,” in other words, the possibility for a person to use the system (Huang et al., 2019).
The successful development of the CAC model has prompted many researchers (Kim et al., 2013; Li, 2013; Lin, 2014) to employ it to investigate users’ attitudes toward information systems (Huang et al., 2019), therefore in this study, CAC model is used to measure users’ attitudes, the dependent variable. The formation of attitudes about news is a process that involves the individual’s perception, understanding, and evaluation of the news story, as well as their emotional reactions and behavioral tendencies in response to the story. How news is presented can have a significant impact on the formation of audience attitudes. In the case of COVID-19, machine-generated news can spread pandemic information with complex, visualized, and dynamic symbolic narrative characteristics. Different characteristics of the information sequences may affect audience attitudes toward pandemic news.
This paper proposes several hypotheses. Hypothesis 1 suggests that the journalist’s identity (robot vs. human) significantly affects individuals’ attitudes toward pandemic news. Specifically, the hypothesis posits that the writer’s identity will have a significant effect on the audience’s content rating (H1a), format perception (H1b), affection (H1c), and conation (H1d) regarding pandemic news. Hypothesis 2 asserts that news format (video vs. non-video) significantly affects the attitudes toward pandemic news. The hypothesis states that news format will have a significant effect on the audience’s content rating (H2a), format perception (H2b), affection (H2c), and conation (H2d) regarding pandemic news. Hypothesis 3 maintains that the interaction between journalists and news formats significantly affects the audience attitudes toward pandemic news. Specifically, the hypothesis posits that the interaction between journalists and format in pandemic news coverage will significantly affect audience content ratings (H3a), format perception (H3b), affection (H3c), and conation (H3d).
The identity of the writer can have a significant impact on audience attitudes. Specifically, viewers were more likely to perceive news stories as biased when the anchor was perceived as having political bias. Other research has shown that the format of news stories can also have a significant impact on audience attitudes. Specifically, when news stories were presented in a traditional, text-based format, viewers were more likely to recognize bias. However, the addition of visual elements made it more difficult for viewers to identify bias.
The interaction between journalists and news formats may also play an important role in shaping audience attitudes, the influence of journalists on audience attitudes may vary according to the different news formats, vice versa. Graefe et al. (2018) found that subjects rated articles declared as human written always more favorably, regardless of the actual source, and subjects rated computer-written articles as more credible and higher in journalistic expertise but less readable.
Overall, the CAC model is a useful framework for understanding how attitudes are formed. This model emphasizes the importance of cognitive and emotional components in forming attitudes, including the role of behavioral tendencies in understanding how attitudes are transformed into actions. Therefore, this model is very suitable for describing users’ feelings and experiences of machine video news, in order to deeply observe the influencing mechanisms of various factors. Therefore, the specific assumptions proposed in this article help us understand how different factors such as journalist identity and news format affect audience attitudes towards epidemic news. By exploring these assumptions, it is possible to better understand how news producers effectively communicate information related to the epidemic to the public.
ANCOVA analysis method and its application
In the research of machine video news, the application status of ANCOVA (Analysis of Covariance) analysis method has attracted much attention. ANCOVA is a statistical analysis method that combines analysis of variance (ANOVA) and regression analysis to compare mean differences between different groups and control for the impact of one or more covariates.
In recent years, many scholars have applied the ANCOVA method to machine video news research to explore the effects of different factors on audience cognition, emotion, and behavior. Using ANCOVA and interaction analysis to explore the impact of different factors on audience reactions and evaluations of video news, and to analyze whether there is an interactive effect between these factors. For example, Liu (2020) used the ANCOVA method to compare the impact of different types of advertising on audience purchase intention. After controlling for other factors, they found that different types of advertising had significant differences in purchase intention.
In summary, the ANCOVA analysis method is widely used in machine video news research, helping researchers gain a deeper understanding of the effects of different factors on audience cognition, emotions, and behavior. And through interaction analysis, we can gain a deeper understanding of the impact of different factors on machine video news, better design and optimize machine generated video news, in order to improve audience satisfaction and user experience. This article uses this method to study and analyze the driving factors of attitude towards machine video news based on the CAC model, which can more accurately test the relationship between various factors and provide strong support for improving the quality and user experience of machine video news.
Research design
The primary purpose of this study is to explore the audience’s attitude towards machine video news. As mentioned above, the audience’s attitude towards machine news and non-machine news is different, as is the attitude towards video news and non-video news. Therefore, the study aimed to examine the effect of journalists’ identities and news formats on audience attitudes toward pandemic news, specifically their content rating, format perception, affection, and conation. The authors of this study conducted an online experiment using a 2 × 2 (robot journalists vs. human journalists; video vs. non-video news formats) factorial design.
Before the participants were exposure to news, they completed a questionnaire that covered control variables, such as gender, educational background, occupation, income, media literacy, news exposure duration, and understanding of AI.
Next, the participants were randomly divided into four groups and exposed to different types of news stories. Each participant was assigned a computer-generated random number (1–100) and divided into four groups (25 participants each group) in descending order. Group 1 was exposed to video news generated by robot journalists; Group 2 was exposed to non-video (text, images, etc.) news generated by robot journalists; Group 3 was exposed to video news produced by humans; and Group 4 was exposed to non-video news produced by humans. The experimental process for each subject lasted about 10 minutes.
Experimental questionnaire measurement scale.
In the cognitive aspect, measuring individuals’ ratings of news content and their perception of news format is crucial. By evaluating participants' evaluation and perception of news content, one can understand their cognitive effects on news reporting, as well as their views on the accuracy and credibility of the report. At the same time, by measuring participants’ perception of news formats, the impact of different forms of news reporting on audience communication and information reception can be revealed. The measurement of the emotional part is to understand the participants' emotional experience when exposed to news. By evaluating the emotions felt by participants when watching or reading news reports, such as joy, sadness, anger, or anxiety, the impact of news content on emotional states can be revealed. This is crucial for understanding how news reporting can evoke emotional resonance among audiences and the impact of emotions on attitude formation and behavior. The measurement section of behavioral tendencies aims to study participants’ behavioral intentions after exposure to news. By understanding whether participants are willing to share, comment on, or forward news reports, the potential impact of news reports on behavior can be revealed. This result can further illustrate the actual impact of news reporting on audience attitudes and behaviors. Therefore, this article chooses to measure cognitive, emotional, and behavioral variables in order to comprehensively evaluate the impact of journalist identity and news format on audience attitudes towards epidemic news. These measurement variables can better help us understand the audience’s evaluation and perception of news content, emotional experience, and possible behavioral tendencies, thereby gaining a deeper understanding of the audience’s attitude and response to news reporting.
As shown in Table 1, we conducted a series of statistical analyses to evaluate the applicability and reliability of measurement projects. Firstly, we calculated the commonality values corresponding to the research project, and the results showed that these values were all greater than 0.4, indicating that these projects were suitable for the participant sample. In addition, we also calculated the cumulative variance interpretation rate after factor rotation, and the result was 63.979%, exceeding the recommended threshold of 60%, indicating that these factors well represent the measured variable. In further analysis, we used KMO (Kaiser Meyer Olkin) measurements to evaluate the appropriateness of sample size. The results show that the KMO value is 0.795, which is higher than the recommended threshold of 0.5, indicating that the sample size is sufficient to support factor analysis. In addition, we also conducted Bartlett’s test and the results showed a
To ensure validity and reliability, we conducted a pre-experiment with 28 participants to test the selection of experimental materials, the number and arrangement of experimental questions, and the clarity of the questionnaire. The pre-experiment procedure was the same as the formal experiment procedure. The subjects were evenly divided into four groups and completed questionnaires after watching the experimental materials. After the pre-experiment, each participant was asked whether they could understand the experimental materials and all the questions in the questionnaire. According to the results of the pre-experiment, the design of the questionnaire was slightly adjusted, and the design of the experimental procedure was retained.
To avoid bias caused by the number of experimental materials, news sources, news types, and participant biases, we chose two news pieces with the same theme, type, and tone for each experiment. All selected news pieces were related to the recent COVID-19 pandemic and were presented in simplified Chinese.
The pre-experiment focused on two different sources of news: Xinhua News Agency and Tencent News. Headquartered in Beijing and founded in 1937, Xinhua News Agency, the largest news organization in China, is the official press agency for the People’s Republic of China, and is the representative of the mainstream media in China, whose website is known as “the most influential news website in China.” Tencent News, as a representative of the commercial media in China, is a popular online news platform where users can access and share news stories from various sources in China. In 2010, the first version of Tencent News mobile application was put on the Apple Store, making it one of the earliest news media to launch client-side products in China. The reason for choosing Xinhua News Agency is that it used Xinhua Zhiyun Epidemic Reporting Robot to report epidemic news professionally during COVID-19, which is in line with the material selection of this study. The reason for choosing Tencent News is that it updated the pandemic news in real time 24 hours during COVID-19, and became an important news source for users to obtain the pandemic information at any time.
The pre-experiment included two types of social news related to the pandemic, which were chosen based on emotional valence. One was positive, containing news about successful vaccination drives, or medical breakthroughs in the fight against COVID-19. The other was negative, containing news about rising COVID-19 cases, or the impact of the pandemic on people’s livelihoods.
The pre-experiment investigated how different tones of news affect people’s emotional and behavioral responses to COVID-19 coverage. Thus, the pre-experiment included two types of news based on their tone. One type was neutral and presented the facts without any bias or emotional messaging. The other had a strong emotional tone, either positive or negative, to evoke an emotional response from the participants.
Following the pre-experiment, a formal experiment was conducted with 100 participants, which included an equal number of male and female participants. None of the individuals were informed of the identity of the journalist or the source of the news. All pieces of machine-generated video news were produced by Xinhua Zhiyun Pandemic Reporting Robot. The formal experiment involved two pieces of social news related to the pandemic with the same theme, type, and tone from Xinhua News Agency and Tencent News. The news pieces were presented to the participants randomly. After watching each piece of news, participants were asked to complete a questionnaire which included questions about their emotional response to the news (e.g., happy, sad, angry, or anxious), and questions on their behavioral response (e.g., intentions to share or comment on the news).
Therefore, this study investigated the impact of news sources and the tones on individuals’ emotional and behavioral responses related to COVID-19. The experiment included two sources of news (Xinhua News Agency and Tencent News), two types of coverage (positive and negative), and two tones (neutral and emotionally charged). Validity and reliability were tested through a pre-experiment with 28 individuals, and the formal experiment involved 100 participants (46 males and 54 females). The research findings can provide insight into individuals’ reactions to social news related to the pandemic. Further, the results reveal how different factors, such as news source and tone, can affect emotional and behavioral responses.
Data analysis
First, the study examined the correlation between the control variables (i.e., age, gender, education level, and occupation) and the dependent variables (i.e., emotional and behavioral responses to the news). The purpose of the analysis was to ensure that the variables did not have a significant impact on participants’ attitudes and, therefore, could be excluded from further statistical analysis.
The results showed the control variables were not significantly related to the individuals’ attitudes. This implies that the variables were not significant predictors in determining emotional and behavioral responses to the news. Therefore, the aforementioned variables were not considered in the subsequent statistical analysis. After screening the control variables, data normality was tested across all dimensions, to ensure that the data met the requirements of parametric analysis. Normality tests are necessary for parametric analysis to ensure adequate sample size and normal distribution of the data. Normal distribution is necessary for a valid and reliable statistical analysis.
The normality of the data was tested across all dimensions. Analysis revealed the data was normally distributed, and the data variance was homogeneous. These results suggest the data met the requirements of parametric analysis, and the statistical analysis was valid and reliable. This further indicates the sample size was appropriate and the data distribution met the standard requirements of parametric analysis. Homogeneity of the variance was also explored, confirming the test was not significant, and the variance was homogeneous. Thus, the data variance in each group was similar, indicating cross comparability.
Third, SPSS was utilized to perform statistical analysis of the data. A variety of statistical techniques were employed to evaluate relationships between the variables and the participants’ emotional and behavioral responses. A descriptive statistical analysis was carried out, which involved calculating basic statistical characteristics such as frequency distribution, mean, standard deviation, and range. The analysis was used to summarize the data and provide information about the characteristics of the sample. This was an important step in evaluating the data before proceeding with inferential statistical analysis.
Next, the researchers performed analysis of covariance (ANCOVA) to test whether the participants’ basic personal attributes would affect the results. ANCOVA is a statistical technique that measures whether the association between the independent variables and the dependent variable is significantly different after controlling for the effects of the covariate. This analysis determined whether the personal attributes of the individuals had a significant impact on their emotional and behavioral responses.
After conducting ANCOVA, the researchers performed a bivariate analysis of variance to study whether the two variables (the journalist and the news format) and their interaction would significantly affect the observed variable, namely the participants’ attitudes. The inter-subject effect test was carried out to measure the effect size and significance of the researcher’s variables on the dependent variable. The analysis determined whether the journalist or news format had a significant impact on emotional and behavioral responses to social news related to the pandemic.
In summary, this article checks and verifies key operations in the experiment according to the following steps to ensure the credibility of the entire experiment: (1) Participants filled out a questionnaire before being exposed to news, covering control variables such as gender, educational background, occupation, income, media literacy, news exposure time, and understanding of artificial intelligence. This step ensures that the background information and features of the participants are similar between the experimental groups. (2) Ensure that the experimental participants are randomly divided into four groups and exposed to different types of news stories: the first group is exposed to video news produced by robot journalists, the second group is exposed to non-video news (text, images, etc.) generated by robot journalists, the third group is exposed to video news produced by humans, and the fourth group is exposed to non-video news produced by humans. By randomly grouping, the possibility of bias and interference in experimental results can be reduced. (3) After reading or watching news reports, participants completed the experimental questionnaire. The questionnaire is selected from existing Chinese and foreign literature and measures variables in cognitive, emotional, and behavioral aspects. The questionnaire has undergone pre experimental testing to ensure the clarity and effectiveness of the questions.
Based on the above methods, this study investigated the impact of news sources and tone on individuals’ emotional and behavioral responses related to COVID-19. The experiment included two news sources (Xinhua News Agency and Tencent News), two types of reporting (positive and negative), and two tones (neutral and emotional). The effectiveness and reliability were tested through a preliminary experiment on 28 individuals, with a formal experiment involving 100 participants (46 males and 54 females). The research results can provide a deeper understanding of individuals' reactions to social news related to the epidemic. In addition, the research findings reveal how different factors such as news sources and tone affect emotional and behavioral responses.
Research findings
Descriptive statistics (Machine-generated video news).
Analysis of covariance of inter-group differences.
The significant values of the journalist factor and the format factor in the participant ratings of news content were 0.000 and 0.002, respectively. In other words, both factors significantly affect news content ratings. Therefore, H1a and H2a have been validated, while H3a does not appear significant. The participants’ ratings of machine-generated news content were significantly higher than human journalist-produced news content (
News format significantly affected the individuals’ perception of news (
The participants’ affection for news was significantly affected by the journalist and news format, with significance values of 0.000 and 0.030, respectively. Therefore, H1c and H2c have been validated. Concerning the effect of the journalist factor, the participants affection for machine-generated video news was significantly more positive than human-produced video news (
The participants’ conation toward news was significantly affected by the interaction between the journalist factor and the format factor (
With regard to the effect of the journalist factor, the participants’ conation toward machine-generated video news was significantly more positive than human-produced video news (
In relation to the participants’ attitudes toward the news, both journalist (
According to recent studies, individuals’ reception of machine-generated video news during a public health event is significantly more positive than their response to human-produced news. This is because machine-generated videos provide valuable information, reliable sources, and clear data guidance that are essential in times of crisis. Reports suggest that the format of news plays a key role in shaping perceptions. Formats not only shape the conveyance of meaning, they influence how information is received and understood. Machine-generated news videos, with their ability to rapidly process vast amounts of data, report events more objectively and precisely, free of the subjective biases that may exist in human-produced news. The efficiency and accuracy of machine-generated video news may enhance the effectiveness of information dissemination during public health emergencies. Machine algorithms can quickly generate videos that accurately reflect the current situation, enabling health professionals and the public to make informed decisions. Additionally, machine-generated videos can be used as a tool for public education and information campaigns, ensuring accurate and timely information reaches a wider audience.
While machine-generated videos offer potential advantages, they should not replace human reporting entirely. Human journalists contribute a unique perspective and context to news reporting. Moreover, they can delve deeper into complex issues, adding enriched perspectives to reports. Interestingly, studies indicate that machine-generated and human-made news videos received higher content ratings, resulting in more positive attitudes from participants, regardless of the journalist type. However, no significant disparities were found in the perception of news formats written by human journalists or non-video news produced by various types of writers.
Therefore, machine-generated video news is a valuable tool for reliable information and insight during public health emergencies. It enhances the efficiency of information dissemination and provides objective, accurate reporting. Machine and human-generated news videos gained high ratings and were positively received by audiences. It is vital to continue improving and exploring the use of machine-generated videos to ensure accurate and timely information is publicly available.
Discussion
Valuable information, reliable information sources, and clear data guidance are indispensable for major public health events (Li & Zhang, 2021). Individuals’ attitudes toward machine-generated video pandemic news is significantly more positive than toward human-produced news. First, format not only affects the method of expression, it also influence show meaning is formed (Lin, 2007).
The immense amount of data generated in the spread of an infectious disease includes time, location, relationship and other aspects in addition to numbers. Visualized pandemic data constitutes the first channel for individuals to learn about the pandemic news in a timely manner (Wang & Xu, 2020). Machine-generated news driven by intelligent technologies differs in physical structure and technical characteristics from its human counterparts. Pandemic data constituted the first channel for the public to learn about the pandemic, command numerous news materials, and fully use diverse symbols and presentation methods to complete the generation of narratives through data visualization.
As a form of enhanced data visualization, data storytelling improves information integrity and systematic perspective through highly intuitive, diverse, and dynamic presentation of data. Visual charts perform better than texts in presenting information and reducing potential biases in textual narratives (McCaffery et al., 2012). Machine-generated news usually involves scientific visual processes and intuitive, eye-catching visuals for narrative design and communication, promoting the conversion of pandemic data into composite, multi-dimensional, dynamic narratives (Ma, 2021). With pandemic information taken into account, interactive data maps, time series diagrams, spatio-temporal cube diagrams, and other visualization tools convey information through different shapes, colors and graphics, achieving visual presentation of different levels and structures.
Second, the public establishes ties with news events through media information. The media exerts influence through affection, prompting people to act. Facilitated by news reporting, public affection will evolve into affective resonance, achieving the convergence and assimilation of collective affection.
Reporting major public health events is crucial for informing individuals and communities of the situation, as well as guiding attitudes and behaviors. However, the emotional impact of epidemic news may be overwhelming, leading to changes in attitudes and behaviors. During the recent pandemic, public anxiety has intensified, and relevant news reports may exacerbate this emotional state. The emotional perception of epidemic news precedes the formation of online public opinion, indicating that during times of crisis, the public is more likely to react emotionally to reports. This emotional response can lead to the loss of rational judgment, which may lead to negative online public opinion and negative social impact.
From the research in this article, it can be seen that the journalist factor and news format factor have a significant impact on the audience’s emotional and behavioral responses. Machine generated news content has a significant advantage in viewership compared to human journalist generated news content, and participants also rated automated video news content higher than machine generated non video news. In addition, news format has a significant impact on individuals’ perception and liking for news, with participants having higher levels of cognition and liking for video news than non-video news. Research has also shown that there is an interaction between journalist factors and news format factors, which has a significant impact on news cognition. Machine generated video news is significantly better in terms of cognition than non-video news, and the audience’s cognition of machine generated video news is also higher than that of human generated video reports. However, there was no significant difference in the perception of non-video news between journalist factors and news format factors. And the factors of journalists and news format have a significant impact on the audience’s attitude towards the news. The overall attitude of the audience towards video news produced by machines is more positive, while there is no significant difference in attitude towards non video news produced by machines or humans. Regarding news formats, the audience’s overall attitude towards video news is also more positive than non-video news.
According to recent research, machine generated video news has received a more positive response in public health events, mainly because they can provide valuable information, reliable sources, and clear data guidance, which helps decision-making during crisis periods. Machine generated video news has high efficiency and accuracy, which can improve the effectiveness of information dissemination in public health emergencies and become an important tool for public education and information promotion activities. However, despite the potential advantages of machine generated video news, it cannot completely replace manual reporting. Human journalists can provide unique perspectives and the ability to delve into complex problems, adding rich content to their reports. Interestingly, regardless of the type of journalist, both machine generated and manually produced news videos have been positively welcomed by the audience.
In summary, machine generated video news is a valuable tool in providing reliable information and insights. It improves the efficiency of information dissemination and provides objective and accurate reports. Both machine and artificially generated news videos have been welcomed by audiences, and machine generated news can alleviate negative impacts by guiding individuals to interpret and perceive epidemic data. The gap between epidemic data and public health information acquisition can be bridged through machine generated news, which will provide information to the public in a more objective and factual manner. This communication strategy can to some extent reduce digital anxiety and cognitive impairment when dealing with complex public health events. In addition, machine generated news can provide more objective and accurate reporting of epidemic events. Unlike human journalists, machine algorithms are less likely to be influenced by biases or personal perspectives, leading to more objective and accurate reporting. This helps reduce unnecessary panic or anxiety among the public and promotes rational decision-making in crises. However, there is still a need for continuous improvement and exploration of the use of machine generated videos to ensure accurate and timely dissemination of information.
Conclusion
This paper examines the impact of journalists (robots vs. human) and news formats (video vs. non-video) on audience attitudes toward pandemic news. Using the CAC model and a 2 × 2 factorial online experiment method, the results show that journalists and formats significantly impact perceptions of pandemic news. Compared to news written by human journalists, the audience’s attitude toward machine-generated video news is more positive, highlighting the potential of machine-generated news to break down digital anxiety barriers and bolster social stability during public health emergencies.
Machine-generated news can play a pivotal role in guiding public perception and interpretation of pandemic events. It can reduce the emotional impact of news and provide accurate, objective, and timely reporting to properly inform the public. By reducing emotional restraints and cognitive barriers, machine-generated news can foster a more informed and rational public response toward public health events. In view of the research results of this paper, we believe that in the face of major emergencies (such as pandemic news), machine-generated news should be used as much as possible, so that the audience’s negative emotions can be weakened to a certain extent.
COVID-19 has impacted the world in unprecedented ways, and media coverage has played a significant role in shaping public opinion and policies. However, the continuous stream of pandemic information has also interfered with people’s perception of facts, leading to extreme emotions and hindering pandemic prevention and control. Thus, the media must report pandemic news rationally and eliminate extreme public moods.
Machine-generated pandemic news has the potential to play a positive role in alleviating public anxiety and maintaining social stability. News generated by machines can create a pandemic data chain to break down data silos and enhance the value of pandemic data. Machine algorithms can also quickly process large amounts of data and generate videos that accurately reflect the situation, enabling individuals to make informed decisions.
The global epidemic has affected the media sector and social information systems. Now, it is necessary to consider how to regulate the relationship between media technology, information dissemination, and audiences. In the current era of high-speed information, technological progress has led to an exponential growth in the momentum of information dissemination. Therefore, identifying the driving factors of attitudes towards machine video news, deeply understanding the impact of different factors on machine video news and the relationship between dissemination and audience, is crucial for controlling the spread of fake news and incorrect information, effectively avoiding the panic caused by fake news and false information and leading to social instability.
To ensure the media maintains its credibility and plays a key role in reporting pandemic news accurately, a regulatory framework that focuses on media ethics, transparency, and accountability must be established. This will guarantee media technology is used for the public good, and the spread of misinformation will be limited.
In addition, with the continuous development of natural language generation (NLG) technology, the application of machine-generated news in finance, business, and other fields will continue to mature. In the future research, besides audience attitude, we should also focus on such issues as readers’ trust in machine-generated news, depth of machine-generated news content, copyright, and infringement.
In conclusion, machine-generated pandemic news can shape public opinion and promote pandemic prevention and control. However, irregulating media technology is crucial to ensure transparency, accountability, and ethical information use. Through regulatory frameworks and improved dissemination of factual information, media can foster rational affective communities, critical thinking, and a better understanding of global pandemics.
Footnotes
Declaration of conflicting interests
Funding
ORCID iD
