Media frames of artificial intelligence
A comparative study between USA and China
Michelle Wartiainen
Bachelor thesis
Department of Government Uppsala University Spring 2020
Supervisor: Alexandra Segerberg Words: 11 041
Abstract
This thesis studies the frequency in which certain views regarding artificial intelligence appear in American and Chinese newspapers. Given the general lack of studies focusing on media frames in the world’s two leading AI superpowers, this thesis contributes to the research on how AI is covered, and how it might relate to the international context. By examining survey data and previous literature, key frames are identified which can be assumed to affect the public’s perception of artificial intelligence. The analytical tool was used in a small-scale quantitative content analysis of 96 newspaper articles. If brief, the results show that the most frequent views of artificial intelligence is expressed in geopolitical, economic, and cultural terms. American newspapers have a higher mention of frames relating to ethics, discrimination and accountability. Furthermore, the results indicate that AI industry products are not as commonly mentioned as previous research have shown. Few articles mentioned artificial intelligence from a dystopian or utopian standpoint. The findings suggest that the reporting about artificial intelligence is not as focused on the AI technologies, but rather how AI as an adopted concept will be a tool, used for national strategies, and hence relevant to include when conducting future studies on artificial intelligence in media.
Keywords: Artificial intelligence, China, America, The United States, The Republic of China, Framing, Frames, Narratives, Media, Media framing, Newspapers, AI arms race,
Innehållsförteckning
1. Introduction ... 2
1.1. Aim & Research questions ... 3
1.2. Outline... 4
2. Literature review and theoretical framework ... 5
2.1. History of AI in society & politics ... 5
2.2. AI and National Strategies ... 6
2.3. Artificial Intelligence in Media ... 7
2.4. Previous Research ... 8
2.5. Theoretical assumptions ... 10
3. Research design & methodology ... 11
3.1. Framing analysis ... 11
3.2. Quantitative content analysis (small-scale) ... 12
3.3. Choice of material: Online news articles ... 12
3.4. Analytical tool ... 14
3.5. Codebook ... 16
3.6. Practical implementation ... 17
3.7. Limitations ... 17
4. Results & analysis ... 19
4.1. Overall results ... 19
4.2. Geopolitics ... 21
4.3. Economic development, competitiveness & accountability ... 22
4.3.1. Economic development & competitiveness ... 23
4.3.2. Accountability... 25
4.4. Societal challenges... 27
4.4.1. Societal challenges ... 27
4.4.2. Jobs ... 27
4.4.3. Scientific/technological uncertainty... 28
4.4.4. Morality/Ethics & Discrimination ... 29
4.4.5. Sensationalist terms ... 30
4.5. Industry products ... 30
5. Concluding discussion & remarks... 32
6. Bibliography ... 34
1. Introduction
“We have been imagining intelligent machines since long before we could build them. As artificial intelligence and robotics begin to fulfil their promise, they therefore arrive pre-loaded with meaning, sparking associations — and media attention — disproportionate to their capacities” (Leverhulme, 2018).
In recent years there has been a rapid improvement of artificial intelligence (AI) technologies due to a rapid increase in available data, algorithms, and processing power. Machine-learning techniques have developed to a point where they are capable of resembling close to, or better than, human-level interaction in confined scenarios (Schwab, 2018, p. 305). In response to the growing AI possibilities, policymakers have presented major political initiatives in order to be at the forefront of the AI and data revolution. The UK government announced the Artificial Intelligence and Data Grand Challenge in May of 2018, which was an effort ‘To put the UK at the forefront of the AI and data revolution’ (Gov.uk, 2018). In 2017, President Vladimir Putin claimed that ‘Artificial intelligence is the future, not only for Russia but for all humankind. Whoever becomes the leader in this sphere will become the ruler of the world’ (Brennen et al., 2018). There is no doubt that artificial intelligence is changing the
international arena in fundamental ways, also changing the way in which countries invest in tech and form their strategies on AI policy.
While policymakers and technologists have begun to discuss AI and the applications of machine learning more frequently, public opinion has not yet shaped much of these conversations (Zhang et al., 2019). As AI spreads into areas of public life through new products, major research initiatives, and decision-making, scholars have pointed to the need of improved understanding of how technical research and expert views are translated into the public (Brennen et al., 2018). Media have played a vital role in shaping ideas, and
The United States and China has been identified as the two leading, and highest ranked countries in the development of artificial intelligence (Tortoise, 2019). Although there are studies that have been conducting framing analysis of AI narratives in media (Brennen et al., 2018; Vergeer, 2020), few studies have assessed the medias in the two leading AI
countries, China and the US. In 2019, Ding and Kong conducted a cross-cultural study that looked at AI development in China and the US. While this study focused on the two countries’ strategic areas of interest in AI products and techniques, e.g., driverless cars, machine, and computing, the study did not cover different framings and narratives in media coverage.
Mainstream media coverage is a key space for understanding public opinion and discussion. It is also a way to investigate how states might form strategies. By recognizing and mapping narratives in public media, one might also realize how media influences the development and adoption of AI technologies (Leverhulme, 2019).
1.1. Aim & Research questions
Research shows that media framing and messaging can form public understanding and opinion on societal issues, providing the public with resources to make sense of, and address pressing public problems. Yet, there are few systematic analyses of how American and Chinese media are covering topics on AI. This thesis aims to fill a gap in the research about media coverage on AI in the United States and the People’s Republic of China. Building on previous research that has identified various media narratives (also referred to as frames), this thesis will investigate which narratives that emerge in American and Chinese news coverage and briefly analyze how these relate to the scholarly debate on American and Chinese strategic priorities in AI development. By conducting a framing analysis on the different media narratives, it will be possible to identify whether any particular perspective is more dominant in the debate, and why. Furthermore, in the light of AI as strategic means in
Following research questions guide this thesis:
I. How do newspapers in the United States and China report about artificial intelligence? A. What frames are associated with artificial intelligence in the articles?
B. What frames are the most prominent among the articles?
II. How does the coverage of artificial intelligence, with respect to national strategies, differ between the United States and China?
1.2. Outline
The disposition of the thesis is as follows: In section two, there will be an introduction of AI and AI development, what role it currently has in society and in politics. There will also be a presentation about how AI is used in national strategies, AI as a tool in the global arms race, later to be followed by central theories of media framing. The theoretical framework will be based on an American public opinion survey about artificial intelligence, and previous research on the discussion and framing of AI. In section three, the research design, material, and methodology will be discussed, together with the construction of the codebook and
2. Literature review and theoretical framework
Following section aims to illustrate the history of AI, and the general academic debate on AI and society, in order to identify the contexts in which artificial intelligence is found, and what this might mean for national strategies. This will be combined with theories on narrative and previous research, laying the basis for my analytical tool used in this study.
2.1. History of AI in society & politics
Artificial Intelligence (AI) covers a wide range of concepts and ideas, and since the field is constantly developing, AI is also hard to define. However, one well used definition is “[The automation of] activities that we associate with human thinking, activities such as decision-making, problem solving, learning” (Bellman, 1978).
AI can be traced back to the 1950’s and ‘60s when it was first theoreticized, but the most subsequent breakthroughs came during the 21st century due to the development of data collecting and aggregation, algorithms, and computing power. An important milestone in AI was reached in 2016, when Google’s AlphaGo computer program defeated the nine-dan professional Lee Sedol at the game of Go, which previously had been considered as unsolvable by machines. This symbolized that it is indeed possible for machines to partly think like humans, and even exceed human capabilities (McKinzey, 2018). In contemporary society, AI and machine learning is often used to process large volumes of data (big data). For instance, many financial market stock trades are programmed by AI-based systems (O’Leary, 2013, p. 96 & 97).
There are constant advances made within the technological field. The adaptation to emerging tech is growing rapidly in various sectors such as health care, manufacturing, finance. According to McKinsey, global venture capital funding of the AI applications market has risen from $589 million in 2012, to over $5 billion in 2016, a number that is estimated to reach $127 billion by 2025 (McKinsey, 2018).
how we work, live, and relate to each other. The complexity, and scope of this revolution is not predetermined, and will require a collected and comprehensive response that needs to involve all stakeholders on the global arena, stretching from civil society, institutions and business, to academia and states (Schwab, 2016, p. 204). Up until this point, scholars have recognized AI as a “wicked problem”, meaning a class of social policy problems for which traditional methods of resolution fail, with no possibility to formulate a definite strategy (Gruetzemacher, 2018). Understanding the ambiguity in policy making towards AI is an important step in realizing the importance of studying how public perceptions and media might shape this debate.
2.2. AI and national strategies
The United States and China have risen to become two leading AI superpowers. Since 2016 and the victory of AlphaGo, China has continuously presented long-term plans and initiatives, such as the Communist Party of China’s 13th Five-Year Plan for Economic and Social
Development to incorporate and develop AI technologies into all sectors (Ding et al., 2019;
Borowiec, 2016). In 2017, the State Council of China presented its ambitious vision of turning the country into the global leader of AI in 2025. AI would become an economic driving force in China, transforming the country to become an AI innovation center in 2025. Artificial intelligence is an omnipresent technology for China, and the tendency to foster it is constantly evolving (Ding et al., 2019). As of now, China does not have the same kind of AI ecosystem as the US. Although China’s internet giants are working on integrating cutting edge technology in many everyday products, generating huge volumes of data from citizens useful when “training” AI systems, professionals estimate that China needs to focus on bolstering its capacity for innovation in order to stay on the forefront in the field of AI and maximize the economic potential of the techniques (McKinsey, 2017).
Preparing, 2016). In 2019, President Trump announced the American AI initiative and their strategy on artificial intelligence, collaborating with the private sector, academia, the public and international partners. The five pillars of the initiative is (1) investment in AI research and development, (2) unleashing AI resources, (3) removing barriers for AI innovation, (4) train AI-ready workforce, and (5) promote an international environment supportive of American innovation and the responsible use of AI (Whitehouse, 2019).
According to Mikael Weissmann (2015), China’s foremost policy objective is
domestic political stability in order to guarantee the survival of one-party rule. The stability is dependent on two factors; continued economic growth and nationalism. Weissman claims that the foreign policy is closely connected to the Chinese self-perception. An important turning point for their identity was the 2008 financial crisis that severely affected the Western economy at a time of Chinese economic success, which resulted in a push of confidence for China to pursue a far more assertive and active stance on the international arena. Under the rule of president Xi Jinping, China will not be a status quo power, but rather a revisionist power intending to remodel the global order (Weissmann, 2015). The rapid economic rise of China, as well as their increased influence on the global stage, has resulted in many scholars trying to understand and analyze China’s motives behind their active AI strategies.
The United States has experienced different approaches towards artificial intelligence with the current administration. Although the Trump administration is now emphasizing an active strategy of America being a leader in AI, the US is only ranked as number 13 when it comes to government AI strategy, while China holds the first place (Tortoise, 2019). This ambiguousness complicates the ability to foresee long term strategies for the US in policymaking when it comes to AI. However, other powerful actors, such as the Silicon Valley tech hub does indeed strive for innovation and has in many ways become an institution of its own with important power and influence that is always not dependent on state support.
2.3. Artificial intelligence in media
private sectors, and surfacing alternative futures to enable debates about these, however, they can also create false perceptions and expectations that are difficult to overturn or change. Depending on whether the user is closely engaged with techniques, narratives might affect confidence in, and perceptions of a technique, and also the confidence and perception towards those who are developing, promoting and opposing them. The theory about narratives
affecting public confidence can also be applied when discussing AI - certain narratives can potentially contribute to a misinformed debate, creating consequences for AI research,
regulation, funding, and reception. By learning about, and identifying narratives, it is possible to widen the spectrum of authors and protagonists amongst AI narratives, creating space for more public dialogues. It is therefore of interest to investigate how media might relate to both public opinions and policy strategies.
2.4. Previous studies on AI
Since artificial intelligence is an emerging technology that is developing constantly, there has been little attention focusing on media coverage, and media framing effects in the literature (Obozinstev, 2018). In 2016, Fast and Horvitz attempted to capture the long-term trend in public debate about artificial intelligence, focusing on views expressed in the New York Times over a 30-year period. By investigating media coverage over time, the authors discerned public concerns and hopes about AI, what ideas were associated with the technologies, and if this changed over the 30-year period.
Since then, there has been research conducted (Obozintsev, 2018; Ding et al. 2018; Brennen et al., 2018; Leigh Star, 1995), with the common denominator focusing on media coverage. Susan Leigh Star (1995) highlighted how American media were presenting AI in terms of sensationalism, often exaggerating the potential and consequences of AI. Brennen, Howard, and Nielsen analyzed eight months of reporting of artificial intelligence by six mainstream news outlets in the UK in order to capture how AI was emerging in the public life. The authors used a mixed-method analysis which revealed that a majority of the articles were pegged to industry concerns, and a large number of products that originated from the industry. By analyzing mass media outlets, the authors identified three frames that were conceptualized in media coverage. The majority of the media was framed around industry
products that include artificial intelligence, such as smartphones, cars, running shoes, or more
controversial products such as sex robots. The second frame was identified as economics and
something that would transform the economic and industrial life, a “fourth industrial
revolution”. This study also recognized aspects of how actors are heralding a new geopolitical order. The third frame was categorized as ethics, discrimination and killer robots, and the findings showed that the ethics of artificial intelligence was one of the most common themes across the corpus.
Frames can thus provide explanations for “how various actors in society define
science-related issues in politically strategic ways, how journalists from various beats selectively cover the issues, and how diverse publics differentially perceive, understand, and participate in these debates” (Nisbet, 2009c, pp. 51). Nisbet’s typology of frames for
scientific issues has previously been applied to studies regarding AI (Obozintsev, 2018). The eight frames that, according to Nisbet, consistently appear across policy debates are (1) social progress, (2) economic development/competitiveness, (3) morality/ethics, (4)
scientific/technical uncertainty, (5) pandora’s box/Frankenstein’s monster, (6) public accountability/governance, (7) middle way/alternative path and (8) conflict/strategy (see appendix A2). In contrast to Obozintev (2018), this thesis will not use these exact frames in the analysis - instead there will be a selection amongst the frames, partly since these Nibset’s typology could not be applied as mutually exclusive, and secondly, because all frames would not be relevant when operationalizing media’s framing of AI.
Ding and Kong (2018) have conducted a study about how Chinese and American newspapers constructed artificial intelligence. By identifying, and comparing the countries different strategic interests of the two countries, they were able to dissect how China and the United States prioritize their strategic paths to develop their positions as leaders in artificial intelligence. The study concluded that the United States and China have different strategic areas of interest in AI concerning techniques such as, computing, internet, pattern recognition, driverless vehicle and big data (Ding et al., 2018).
the development of AI, while a smaller minority (22%) somewhat or strongly opposed it. Demographic characteristics such as education, income, knowledge about computing and gender, accounted for substantial variation in support for developing AI. A majority of Americans (82%) believe that robots and/or AI should be carefully managed. The study findings highlighted government challenges such as surveillance and civil liberties, the right to privacy, cyber-attacks and preventing AI technologies from spreading fake and harmful content online (Zhang el al., 2019).
2.5. Theoretical assumptions
Based on previous research and by putting artificial intelligence into an international context highlighting American and Chinese prerequisites for requiring AI, it is possible to construct the theoretical framework of this thesis.
The Zhang and Dafoe (2019) survey opinion demonstrated that the American public have opinions about how to manage artificial intelligence, as well as a general view of AI. American newspapers could be expected to cover AI from a more critical stance, since the media, opposed to China, is independent from the state. For the results, this might mean that narratives, and frames regarding moral, ethics, and accountability is more prominent.
With regard to Chinese newspapers, based on the communist party’s continuous strive, and measures towards becoming a global leader in AI and economic growth, as described by Kai-Fu Lee (2018), one expected result would be a high mention of economics and innovation in newspapers. Since media is controlled by the government, the newspapers could be expected to have a more “positive” approach towards AI technology.
3. Research design & methodology
Following section is presenting the choices of method in this thesis. The first part describes framing analysis and its relevance for this study, which is followed by an explanation of quantitative content analysis. My selection of material will thereafter be motivated and discussed. After illustrating the codebook, a more thorough description of its implementation will be presented, together with a section discussing limitations of the study.
3.1. Framing analysis
The approach of framing was introduced by Erving Goffman in 1974, and in his ethnographic research, frames were described as allowing individuals to “locate, perceive, identify and label” various topics, issues, and events. According to Goffman, words can be interpreted as triggers, helping individuals negotiate meaning through existing cultural beliefs and
worldviews. Since Goffman’s introduction of frames, researchers have explored and explained how media portrayals shape public views. Frames organize central ideas on a particular issue, and are used by public as “interpretative schema” in order to make sense of, and deliberate on issues. They are also used by policy makers to define political strategies and reach consensus in decisions (Nisbet, 2008). Over the past decades, frames have since been used as a method in several studies that aim to investigate what effects can be seen when the media is using it. Frames affect how citizens form their opinions about institutions, political issues and news and the consensus amongst scientists is that media reporting affects the individual, although there is not a clear causation that clarifies this (Shanahan et al., 2008, p. 116–118; McKay et al., 2011, p. 611).
3.2. Quantitative content analysis (small-scale)
Quantitative content analysis is used in the sense that the analysis is based on comparable data that can be analyzed consequently by using numbers and statistical methods. It can be applied on both written or spoken sources, and quantitative content analysis is often used as a method in political communication research when the ambition is to investigate whether something is occurring or not in the content subject to analysis. (Esaiasson et al. 2014, p. 197). The
strengths of conducting a quantitative content analysis is its sample size, which often results in “a reduction in sampling error and an increased level of confidence in generalizing from our results” (Manheim and Rich 1995, p.186). This study is considered to be a small-scale content analysis. How well it will objectively identify patterns that can be generalizable, will be discussed under practical implementation (3.2).
Performing a quantitative content analysis normally includes a codebook to clarify the variables and the values they might take. A codebook manual will also specify how the values are to be interpreted empirically. Principles and rules are often necessary to define in more ambiguous cases and studies in order to maintain a high reliability throughout the coding process. (Esaiasson et al. 2014, p.200-201).
3.3. Choice of material: Online news articles
The findings described in this thesis derives from a small-scale content analysis of American and Chinese articles published in 2019. The search word criterium for using an online article was “Artificial Intelligence”, which was selected because the term can associate to many technologies, but also capture a larger debate about general opinions about AI. Searching on the word “AI” decreased the search results on the database and would might have risked capturing other phenomenon’s outside the scope of artificial intelligence. I also wanted to select as few words as possible, for the sake of making the process and effective. The chosen material comprises written content related to Artificial Intelligence, including news, features, opinion pieces. I did not exclude any material, except written content that were disconnected completely to AI, for example when mentioning the educational background of a person.
would reach and influence a wide range of individuals. Accessibility on how easily relevant material could be accessed to, and eventual limitations from the database Factiva. Workability would capture how easy the material could be processed keeping reliability in mind.
Television and morning news shows are for example popular in the United States, and would also meet the demand on relevance. However, these types of medias have a disadvantage in meeting the criteria on accessibility and workability.
This study processes material and content that reach a large mass of Americans and Chinese citizen. Four widespread newspapers were chosen, two newspapers for each country. Two articles were randomly chosen per month, meaning there were 24 for each magazine (New York Times 24, Forbes 24, China Daily 24 and People’s Daily 24 - 96 articles in total). The majority of the articles were collected from the physical newspapers, but since Forbes magazine are releasing fewer issues per year, I complemented with online articles from the website. Features and articles that are published on a website, might not reach the same people buying a paper magazine, which is worth taking into consideration. However, in this study the more important focus was to find a wider range of material. The papers chosen are considered to be among the most influential newspapers in each country. In the United States, The New York Times and Forbes was chosen for this study. The New York Times is widely thought upon as an authoritative source when setting the national news agenda (J. Kraus, K. McMahon, D. Rankin, 2006). To complement this, I chose Forbes which is the largest bi-weekly American financial and business magazine, meaning this would spread the readership basis. If there were more resources in this thesis, one could have investigated more
newspapers, television features, and other types of medias.
China Daily (CD) has been described to be “the mouthpiece” of the Communist Party of China, targeting international audiences and also being one of the most cited by foreign media (Scollon, 1998 & Stone, 1994). As for China Daily, and People’s Daily, they both met my chosen criteria of relevance, accessibility, and workability. China Daily is the national English-language newspaper, reaching more than 200 million people worldwide, thus mirroring what strategies China might want to convey to the outside world (China Daily
About us 2020). People’s Daily is also published in both Chinese and English, reaching
3.4. Analytical tool
According to Manheim and Rich (1995, p.185) the creation of a functioning and suitable analytical tool is one of the most time-consuming and challenging part of a quantitative content analysis. Building on previous research, the coding in this study is inspired by different AI frames provided by previous scholars. The analytical tool in this study is based on 16 different variables in order to capture what frames mentioned in the different
newspapers. The variables are inspired by ready-made categories of other scholars (Brennen et al., 2018; Nisbet, 2008) that was presented in section 2.2. Although, this analytical tool is modified, refined and developed to suit this particular research purpose and questions. Following table explains how to interpret the variables in the codebook (3.5);
Table 3.1. Illustration of analytical tool to use when coding for frames
Frame Interpretation of an AI-related issue Geopolitical
changes:
- does the text capture a structural or significant shift in geopolitics, due to development of AI? In what sense? Economic
development:
- are there mentions of an economic investment; market, benefit or risk in terms of AI?
Competitiveness: - are there mentions in terms of conflict and strategy, such as who is winning/losing/leading in developing (the latest) technology/AI?
Governance and accountability:
- are there mentions of research or policy serving public interest or serving special interest; emphasizing issues of control, transparency, participation, responsiveness or ownership?
Societal challenges:
- is AI framed as a societal challenge?
Jobs
opportunities:
- will AI bring more/less job opportunities?
Scientific/ Technical uncertainty:
- are there any debate over proper use of science and expertise amongst stakeholders in AI, what is known versus unknown? Morality/ ethics: - is there a discussion of AI as a matter of right and wrong,
Discrimination: - is AI discussed in terms of, or as a technique to discriminate against actors?
Sensationalist terms:
- is there a need for precaution in the face of possible out-of-control consequences of AI, or is AI a bringer of a futuristic utopian ideal?
Industry products: - is the primary focus in the article to describe or promote an industry product, such as a phone, car, etcetera?
In order to treat the articles in a mutually exclusive way, I also added the category “neither or not mentioned” (often labeled as 0 in the codebook). These will not count as a mention when presenting the results, but is simply a way to illustrate if the topic was not talked about or not. There are also categories labeled as “both/other/mentioned” (often labeled 3). The advantage in using a category concerning “other”, is minimizing the risk not capturing interesting aspects of the debate. Since AI is in constant development, one can assume that new and interesting perspectives will be brought into the light. One important approach has been not to create too many categories in one variable, since valuable information could be lost if it is not possible to tell the aspects apart in the final results.
The process of creating the codebook was conducted in two steps, the first being to test the code book on a small number of articles in order to see if the categories were suitable and mutually exclusive. The second step was to provide the manual and codebook to a test person in order for them to code one article that had been coded by me in advance.
3.5. Codebook
Variable 1: Newspaper
1 - The New York Times (American) 2 - Forbes (American)
3 - China Daily (China) 4 - People’s Daily (China) Economics & Geopolitics
Variable 2: Geopolitical changes
0 - Not mentioned 1 - Economy 2 - Military 3 - Culture 4 - Other/Combined
Variable 3: Economic development
0 - Not mentioned
1 - Promoting economic development 2 - Undermining economic development 3 - Both/other/mentioned
Variable 4: Competitiveness
0 - Not mentioned
1 - Leading role in AI development 2 - Falling behind in AI development 3 - Both/other/mentioned
Variable 5: Governance & Accountability
0 - Not mentioned
1 - Possible to hold actor accountable 2 - Difficult/not possible for accountability 3 - Both/other/mentioned
Societal challenges
Variable 6: Societal challenges
0 - Neither 1 - Yes 2 - No 3 - Both/other/mentioned Variable 7: Jobs 0 - Neither 1 - Jobs lost 2 - Jobs created 3 - Both/other/mentioned
Variable 8: Scientific/Technical Uncertainty
0 - Neither 1 - Yes 2 - No 3 - Both/other/mentioned Variable 9: Morality/Ethics 0 - Neither 1 - Issue 2 - Not an issue 3 - Both/other/mentioned Variable 10: Discrimination 0 - Neither 1 - Yes 2 - No 3 - Both/other/mentioned Sensationalist terms
Variable 11: Sensationalist terms
0 - Neither
1 - Pandoras box, dystopia 2 - Utopian ideal
3 - Both/other/mentioned Industry
Variable 12: Industry product
0 - No 1 - Yes Other
3.6. Practical implementation
In this part I will provide the general procedure of the study, analysis, and the principles used when facing ambiguous situations, since coding should be done in a systematic and
comparable matter when conducting a quantitative content analysis (Esaiasson et al., 2014:200).
To begin with, “artificial intelligence” was the search term that generated the articles, however, not all articles containing the term artificial intelligence were suitable for the analysis. This was for example when only mentioning artificial intelligence as a title of a person’s education or personal profile. This resulted in a few articles being rejected. The year selected was 2019 (January 1st to December 31st) for all newspapers. For each magazine, I systematically chose the first articles that appeared in the search flow from each month, which meant they were published towards the end of the month due to the layout of the database. 96 articles were conducted for coding, and generated a total of 345 views/mentions (see table A1 in appendix). Excel was used to keep track on the initial coding, but also for creating
calculations and illustrations. To maintain a high reliability, it was decided not to code articles that did not explicitly mention, or implied in a strong sense, that a particular view was
expressed. In ambiguous cases, it was coded for “other/both/mentioned” (3) and made short notes on what the issue concerned. That was to avoid subjective judgements on what the argument or themes were. If a view was expressed numerous of times, it was not coded several times, since this would have created misleading results by exaggerating the most frequent findings that could not be treated as how a reader would understand the general content.
Although the ambition was to determine views by explicit mentions or strong implications, there were also situations where personal interpretations were necessary. For example, if an author quoted one opinion to present an oppositional view. In these cases, it was necessary to determine what general sense/tone the author wants to convey.
3.7. Limitations
4. Results & analysis
The following section will present the result of the media analysis in order to identify while discussing general frames and patterns. Quantifiable results are to be illustrated graphically together with quotes taken from analyzed articles, in order to dissect and motivate relevant findings within the different frames. A detailed table with exact encodings per newspaper is attached in the appendix (table A3).
4.1. Overall results
%
Figure 4.1: Average frequency (%) of mentions in media
The overall results of the media analysis show that the newspapers subject to analysis, both in the US and China, are frequently highlighting geopolitical aspects of artificial intelligence, which makes for almost 30 percent of all mentions. American and Chinese newspapers have a similar number of mentions regarding competitiveness (29 for the US and 34 for China), thus mirroring the scholarly debate of innovation, and AI as being prominent as a national strategic area in an international relations context.
For China, the average mentions are higher on three variables: economic development, competitiveness, and jobs. Specific mentions about artificial intelligence promoting economic development made up 18 percent of all the mentions from Chinese newspapers, and 10
economic development only make up 6,15 percent, and 5,1 percent in terms of AI changing the economy in a geopolitical matter. In categories such as societal challenges, accountability and discrimination, the average frequency for the US is higher than China. This will be further discussed later in the analysis.
In terms of discussing industry products, the overall findings show a low frequency of highlighting these topics (see appendix A1), which is interesting due to the findings from previous research that conclude a very high frequency of industry products mentioned in the media.
%
Figure 4.2: Frequency of mentions in American & Chinese newspapers
4.2. Geopolitics
The geopolitical variable is making up almost 22 percent of all mentions in the results. This might be due to the wide scope of the variable. Geopolitics is divided into different
alternatives such as economic changes, military changes, cultural changes and one alternative capturing “other answers” in order to conduct a mutually exclusive coding. Geopolitical changes in the military was only mentioned once in a New York Times article written by the former US marine Lucas Kunce. The author pressed on the need to differentiate AI in war by separating the decision to go to war and what happens on the battlefield during warfare. The article pointed towards how artificial intelligence could change the practical methods of warfare, such as improving robot techniques and minimizing unnecessary harm by letting AI make decisions in critical situations.
“The military's goal in warfare isn't to kill as many people as possible, but to accomplish missions that our country sends us to accomplish while minimizing harm to civilians” (Kunce, 2019; see table A3 (11))
A more addressed category within the geopolitical variable is economy. The first quote is from an article describing how the US-China trade war has enabled China to rethink and reinvest in their techniques (such as AI) in the international arena, the second quote is describing how tech has helped the influence of Chinese companies on the international arena:
“China is not afraid of a trade war, because it can use it to deepen reform and opening-up” (Peng Bo, 2019; see table A5 (61))
“Based on big data, the company could stock 82 percent of goods in advance. It has helped 5,000 Chinese companies go overseas and sell their products to Belt and Road countries and regions” (People’s Daily, 2019; see table A6 (76))
Many coding results labeled as “other/both/mentions” were discussion politics, or a combined view of AI changing the economy, and the cultural way in which we live. This indicates that
politics would in fact have been a valuable category to add into geopolitics, something that
culture. The first one describing AI techniques that can mimic human painting in an
unrecognizable way, which challenges the way in which we view art. The second article was published by Forbes, discussing how artificial intelligence could change the traditional fashion culture.
"When nobody can tell the difference between paintings generated by AI and those of humans, then we can regard AI's work as original" (People’s Daily, 2019; see table A6 (86))
“Just as the world of finance has its own fintech, so in fashion fashtech is slowly being born. Robots will not only dress us up, but also advise us what to wear” (Forbes, 2019; see table A4 (30)
The final category other captures changes in geopolitics that exceeds the categories of
economy, military, and culture. The most common examples that were brought up were often related to politics (example 1) or other fields that could change such as medicine
(example 2).
“By creating these digitally manipulated videos, Google's scientists believe they are learning how to spot deepfakes, which researchers and lawmakers worry could become a new, insidious method for spreading disinformation in the lead-up to the 2020 presidential election” (Metz, 2019; see table A3 (5))
“A firm in South San Francisco that aims to find new drugs by sorting through masses of data. If it succeeds, it will have overturned how drugs get discovered” (D’Onfro, 2019; see table A4 (28))
The major findings in geopolitical changes is thus the frequency mentions of economic, cultural, but partly also political aspects that are brought up in AI technologies. To continue, we will take an in depth look at the following economic and geopolitical variables such as economy, competitiveness, and accountability.
4.3. Economic development, competitiveness & accountability
Figure 4.3: Views regarding geopolitical & economical aspects in American newspapers
Figur 4.4: Views regarding geopolitical & economical aspects in Chinese newspapers
4.3.1. Economic development & competitiveness
Total US: 7.15%
Promoting economic development: 6.15% Undermining economic development: 0%
Total China: 19.4%
Undermining economic development: 0%
The single most common observation in Chinese media is the one implying that
competitiveness in artificial intelligence promotes economic growth and stabilizes China’s position as an economic global leader. Even if the results in this study found could be affected by temporary changes in media content over this particular time period, this observation should make it safe to assume that the view on artificial intelligence as promoting economic development in the region is indeed the most commonly emphasized one in Chinese media, some examples being;
“It also effectively helps Huawei build its digital ecosystem around its AI chip hardware portfolio, which is critical to the localization strategy for customers in China to ensure long-term business continuity in the increasingly dynamic economic environment” (Shenzhen et al., 2019; see table A5 (70))
“In January, the New Orders Index of the Purchasing Managers' Index (PMI) stood at 46.9 percent in China, up 0.3 percentage points from the previous month. The index for high-tech
manufacturing hit 51.6 percent, much higher the industry's general level” (People’s Daily, 2019; see table A6 (76))
There are also mentions of China’s AI development as being world leading, and able to seriously shape the international arena due to AI tech or investments in foreign regions, often with regard to the Belt and Road countries, or African regions.
“Post-2001, China's story has been no longer only about attracting foreign investment. It is also about Chinese companies "going global". China is no longer just the "factory of the world" but also the "investor of the world". The new face of China is represented by Chinese multinational companies as innovators and investors in high-tech industries” (Brahm, 2019; See table A5 (63)) “Huawei's competitive edge in data storage, cloud services and artificial intelligence can be harnessed to hasten Kenya's transition to a knowledge-based economy” (People’s Daily, 2019; see table A6 (77))
“China has agreed to advance high-quality development in tandem with the other BRICS member states (Brazil, Russia, India and South Africa), triggering a new round of industrial revolution featuring biotechnology and artificial intelligence” (Xiaohu, 2019; see table A5 (71))
separate between newspapers, the first quote being from an American newspaper expressing concern over the arms race with China, and the second quote from People’s Daily, describing how the economic and technological trade war (initiated by the US) is hurting not only China and the US, but also the whole world order.
“For American officials, the stakes seem much higher now than in the race with Japan. Most economists estimate China will overtake the United States as the largest economy in 10 to 15 years. And some senior officials in Washington now view China as a steely ideological rival, where the Communist Party aims not only to subjugate citizens but to spread tools of authoritarian control globally -- particularly surveillance, communications and artificial intelligence technology -- and establish military footholds across oceans and mountains” (Wong, 2019; see table A3 (16)) “China insists upon trade liberalization and is continuously lowering its tariff rates…” “... cooperation between the two countries benefits not only their combined population of more than 1.7 billion, but also people in the rest of the world. And confrontation between them would spell disaster for their peoples as well those in the rest of the world”. & “Besides, there is a great difference between China's inclusive and the US' exclusive policies. China has always advocated "cooperation and negotiation", while the US seems hell-bent on treading the path of "conflict and confrontation", as it has not only launched military wars but also triggered trade and technological wars” (Angang 2019; see table A5 (60))
Three key findings can be derived from this section: firstly, China is aiming towards taking a clear lead when it comes to developing and spreading AI tech, and Chinese media is
indicating that China is changing the international arena. American newspapers are also discussing AI in terms of improving the economy, but not to the same high extent as Chinese. Secondly, there are also a high frequency of concern regarding competitiveness (or the AI arms race) between China and the US, American newspapers painting a more vivid scenario of China's’ rampaging as a threat to the world order, and Chinese newspapers pointing out the US as the part who is failing the global community. A third interesting finding in both
countries is that there are no clear mentions of artificial intelligence being an investment risk, and possibly hurting the economy.
4.3.2. Accountability
regarding AI tech that could develop into having clear negative consequences for democracy, for example Deep fakes - which is an AI tech able to digitally manipulate videos (Metz, 2019). Some articles focused more on pushing state-, and international actors to push for accountability.
“So in 2015, at a scientific conference, I organized a debate on this new class of weaponry. Not long afterward, Max Tegmark, who runs M.I.T.'s Future of Life Institute, asked if I'd help him circulate a letter calling for the international community to pass a pre-emptive ban on all autonomous robotic weapons” (Dreifus, 2019; see table A3 (13))
There are only three Chinese articles mentioning accountability regarding AI. Accountability and the government's responsibility towards citizens in terms of AI development is discussed about more as an area of improvement than direct criticism. Following quote is taken from People’s Daily.
“The four-day Party meeting will also mark a new phase for China in terms of improving
economic governance to better facilitate the country's opening process and its integration into the global economy as the country is seeking to attract foreign investors with a more sound and mature economic system, they added” (People’s Daily, 2019; see table A6 (91))
The major finding with regard to accountability is the difference approach towards accountability in the two countries. In China, accountability is seen as an area of
4.4. Societal challenges
%
Figure 4.5: Figure 5: Views on AI in American and Chinese newspapers concerning societal issues - not including the alternative “other/both/mentioned”
Note:
1 - No societal challenges/Societal challenges 2 - Jobs created/Jobs lost
3 - No technological uncertainty/technological uncertainty 4 - No ethical and moral issues/Ethical and moral issues 5 - No discrimination/Discrimination
6 - Utopic/Dystopic
4.4.1. Societal challenges
Societal challenges connected to AI is frequently mentioned in the newspaper articles. When calculating an average measure of our results, we can see that the “negative” approaches on topics such as technical uncertainty, ethics, and discrimination is dominating. As stated previously, this figure is illustrating “positive” and “negative” views. The section will also demonstrate how the countries might differ when it comes to approaching these topics.
4.4.2. Jobs
“On the looming prospects of AI stealing human jobs and causing massive layoffs, Ma challenged the assumption that there would be a job shortage in the future. The invention of electricity gave humans more spare time, so "you can go to a dance party in the evening," argued Ma. Likewise, he believes more powerful AI will give humans "more time to enjoy being human beings” (People’s Daily, 2019; see table A6 (87))
“Chinese authorities recently announced the plan to add 15 new job roles in fields including artificial intelligence and big data” (People’s Daily, 2019; see table A6 (73))
There are also articles pointing out that the labor market will change, but not in the way we might think, such expressed by Jamie Condliffe writing for the New York Times.
“It found that A.I. will be ''a significant factor in the future work lives of relatively well-paid managers, supervisors and analysts,'' including those in relatively technical roles. Perhaps the most surprising finding: Holders of bachelor's degrees would be exposed to A.I. over five times more than workers with only a high school degree” (Condliffe, 2019; see table A3 (6))
These results imply that there is a clear consensus about how the future of the labor market might look, although we can see a slightly higher frequency in mentions of jobs being lost. Interestingly, this topic was frequently mentioned compared to other categories under societal challenges, which will be further discussed later.
4.4.3. Scientific/technological uncertainty
There are a not many expressed concerns when it comes to the uncertainties with AI
technology and eventual problems in controlling it, following quote being from the New York Times.
“The technologies used to create deepfakes is still fairly new and the results are often easy to notice. But the technology is evolving. While the tools used to detect these bogus videos are also evolving, some researchers worry that they won't be able to keep pace” (Metz, 2019; see table A3 (5))
techniques. Overall, the topic of scientific uncertainty was not frequently touched upon as frequent, however, since this category refers to the technological and scientific uncertainties in constructing AI tech, it does not rule out that there might be a larger uncertainty towards the concept of AI.
4.4.4. Morality/ethics & discrimination
Moral and ethics was a fairly discussed topic, making for 8 percent of the US mentions. The New York Times was the far most dominating newspaper discussing AI from an ethical and discriminatory perspective, which is illustrated in figure 2 under “overall results”. Examples of how the newspaper could approach issues of ethics and moral is illustrated with following quotes, the first one being suggestions on how to limit the unethical power Facebook has over American citizen.
“How do you meaningfully punish a company as enormous and profitable as Facebook”? “... a federal technology standards authority … stop processing [data] orders. Individual liability for directors. Mandatory server audits. Statutory rights to compensation for consumers. Safety inspection and licensing prior to scaled release of new features” (Warzel, 2019; see table A3 (20)) “... there are still a lot of challenges facing the technology before it can widely be applied in various industries. The challenges include scientific, technological and also ethical” (Shenzhen et al., 2019; see table A5 (64)).
With regard to discrimination these quotes are both taken from New York Times articles:
“Racism in technology is a serious problem, and I know there are many people who are much smarter than I am who are thinking about how to solve it” (Charlton, 2019; see table A3 (7)) “You can't have machines deciding whether humans live or die. It crosses new territory. Machines don't have our moral compass, our compassion and our emotions. Machines are not moral beings” ... “What makes these (AI weapons) different from previously banned weaponry is their potential to discriminate. You could say, ''Only kill children,'' and then add facial recognition software to the system” (Dreifus, 2019; see table A3 (13))
4.4.5. Sensationalist terms
“When I think about A.I. and weapons development, I don't imagine Skynet, the Terminator, or some other Hollywood dream of killer robots.” … “... technologies from applications of A.I. to augmented reality would save innocent lives and reduce suffering” (Kunce, 2019; see table A3 (11))
The Kunce article is one of the three articles touching upon the major benefits with AI and what implications this will have on society. The following quote describes a quite common view when talking in sensationalist terms.
“... Big Tech companies becoming targets of major antitrust investigations by the nation's federal enforcement agencies, Congress and states. The concern is not only the companies' market power but also how they handle personal data, spread disinformation and amplify partisan divisions. Then, the technology engine was the industrial revolution. Today, it is the digital revolution. Then, the targets of enforcement were the industrial trusts, led by Standard Oil. Today, it is the digital behemoths. Then, the policy response was new antitrust laws -- the Sherman Act of 1890 and the Clayton Act of 1914 -- and aggressive enforcement. Today, the policy response remains to be seen” (Lohr, 2019; see table A3 (3))
Sensationalist terms are not addressed as much as expected when reading about previous research. Although there is a marginal difference in coding when reading articles that express a worry in regard to AI, versus articles using sensationalist terms such as “digital behemoths”.
4.5. Industry products
Table 4.1: Number of articles mentioning industry products in American and Chinese
newspapers American newspapers
Articles Chinese newspaper Articles
The New York Times
0 China Daily 3
Forbes 13 People’s Daily 7
To conclude, China and the US had a similar frequency when it comes to industry products, however, the findings shows that industry mentions were the most prominent in Forbes, who covered products such as Chatbots, Nuclear technology, Self-driving cars, Footwear etc. This result differs from previous research (Brennen et al., 2018), who found that 60% of UK articles looked at were focusing on industry products.
In regard to China, 4 out of 10 articles concerned Huawei in some sense, for example Huawei achieves AI breakthrough (People’s Daily, 2019). The article address that Huawei secured an important step in chip research with its self-developed AI intelligence computing, which could be used in Huawei’s Atlas 900 to boost a computing power program. One Forbes article described how AI could be used in developing fusion as an energy source.
5. Concluding discussion & remarks
The aim of this study has been to contribute in filling an existing research gap on what
narratives and frames are the most prominent in American and Chinese media. This was done by identifying frames found in previous research as well as survey data that is believed to shape public perceptions. Quantifying the results laid the foundation for investigating the most interesting and prominent frames for the selected countries, as well as recognizing general patterns, and how they might relate to previous research, scholarly findings, and theories about national strategies.
Geopolitical and economic frames have shown to be prominent in this study. The findings from Chinese newspapers indicated that AI is framed as a means for economic growth, but also as a tool to maintain the position as a powerful, and influential global leader. The frames can thus be assumed to align with the Chinese self-perception, and their national strategies about maintaining the economic growth in order to stabilize the one-party rule (which was presented in section 2.2). AI is also framed as improving the economic development in American newspapers. The US administration is presenting ambitions AI initiatives, but it ranks low compared to China when it comes to governmental strategies. However, according to the findings in this study, American newspapers are still framing the United States as a leader in AI, frequently highlighting American success-stories in
developing AI. This could be due to a strong driving AI development in the United States that is not entirely dependent on national strategies - Silicon Valley.
One presumed assumption was how American newspapers might have a higher frequency of mentioning topics such as accountability, moral, ethics, and other societal
challenges. This assumption was partly based on the fact that the United States is a democracy with a free press, which can result in newspapers being able to criticize powerful actors, informing the public on pressing issues. Results showed that American newspapers indeed have a higher frequency mention than Chinese media on ethics, discrimination, and
selected countries for this study was chosen because they are both considered as the leading AI countries in the world, and although there is considerable difference such as governmental style, a comparative perspective is still important to assess how the countries view and relate to AI. The newspaper having the highest mentions of ethics, discrimination, and
accountability was the New York Times, which makes it interesting to investigate how
different types of newspapers might differ in the same country. However, this study was more comparative, highlighting how major frames between the US and China can differ, in order to say something about their leading stance, and national strategies connected to AI. Further studies might look into how different medias may frame AI, although such a study would only be applicable in a country as the US, that has a wide range of independent newspapers and press, and not a country with a one-party rule as China.
This thesis is contributing by giving insight to which narratives and frames are
covered in the Chinese and American newspapers, as well as how often and how. By studying articles from 2019, there seem to be a focus in media coverage on AI that is not necessarily emphasized in previous studies. There are few mentions of industry products that were dominating the medias in the UK (Brennen et al., 2018), and few mentions in terms of sensationalism as Leigh (1996) found. The most mentions on artificial intelligence frame AI as a geopolitical aspect, either by confirming that AI will change the economy, culture, and politics in vital way, or framing AI as a tool that countries may use to maintain their position on the international arena, and develop their economies. This seems to highlight an important area when studying AI superpowers. The publics might not be as interested about hearing about what AI products will be able to do. There might be a more acceptant approach towards AI, since this study did not find strong opposite sides of utopic and dystopic discussions. This result can point towards that there is less focus on AI techniques themselves, but rather on how they will be used and what they do for, and bring to the countries, participating in the AI arms race.
6. Bibliography
Artificial Intelligence for the American People. White House.
https://www.whitehouse.gov/ai/executive-order-ai/. Retrieved: 2020-04-12
Barton, Dominic. Woetzel, Jonathan. Seong, Jeongmin. Tian, Qinzheng. (2017). Artificial intellegence: Implications for China. McKinsey Global Institute.
https://www.mckinsey.com/~/media/McKinsey/Featured%20Insights/China/Artificial%20inte lligence%20Implications%20for%20China/MGI-Artificial-intelligence-implications-for-China.ashx. Retrieved: 2020-04-12
Bellman, Richard Ernest 1978, An introduction to artificial intelligence: can computers
think?, Boyd & Fraser Pub. Co, San Francisco.
Borowick, Steven. 2016. AlphaGo seals 4-1 victory over Go grandmaster Lee Sedol. The Guardian.
https://www.theguardian.com/technology/2016/mar/15/googles-alphago-seals-4-1-victory-over-grandmaster-lee-sedol. Retrived: 2020-04-01
Brennen, Scott. J. Howard, Philip. N. Nielsen Kleis, Rasmus. (2018). An Industry-Led Debate: How UK Media Cover Artificial Intelligence. Reuters Institute.
https://reutersinstitute.politics.ox.ac.uk/sites/default/files/2018-12/Brennen_UK_Media_Coverage_of_AI_FINAL.pdf. Retrieved: 2020-04-01 China Daily. About Us.
http://www.chinadaily.com.cn/e/static_e/about. Retrived: 2020-
Craig, Claire. Cave, Stephen. Dihal, Kanta. Dillon, Sarah. Montgomery, Jess. Singler, Beth. Taylor, Lindsay. (2018). Portrayals and perceptions of AI and why they matter. Leverhulme Center for the Future of Intelligence & The Royal Society.
http://lcfi.ac.uk/projects/ai-narratives-and-justice/ai-narratives/. Retrieved: 2020-04-20 Esaiasson, Peter. Gilljam, Mikael. Oscarsson, Henrik. Towns, Ann. Wängnerud, Lena. (2017). Metodpraktikan : konsten att studera samhälle, individ och marknad. Upplaga 5. Sverige: Wolters Kluwer.
Ding, Huiling. Kong, Yeqing. (2019). Constructing Artificial Intelligence in the U.S. and China: A Cross-Cultural, Corpus-Assisted Study. China Media Research (Vol. 15, Issue 1). Edmondson Intercultural Enterprises.
Fast, Ethan. Horvitz, Eric. (2016) Long-term trends in the public perception of artificial
intelligence. Cornell University.
https://arxiv.org/abs/1609.04904. Retrieved: 2020-03-26 Gov.uk. 2018. ‘The Grand Challenges.’
https://www.gov.uk/government/publications/industrial-strategy-
Gruetzemacher, Ross. (2018). Rethinking AI Strategy and Policy as Entangled Super Wicked Problems. AIES '18: Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society. p.122. https://dl.acm.org/doi/10.1145/3278721.3278746. Retrieved: 2020-04-03 Kai-Fu Lee. (2018). AI Superpowers: China, Silicon Valley, and the New World Order. Boston, USA: Houghton Mifflin Harcourt. p. 52
Kraus, Jon. McMahon, Kevin. Rankin, David. (2004). Transformed by Crisis: The Presidency of George W. Bush and American Politics. New York: Palgrave Macmillan.
Maurice Vergeer. (2020). Artificial Intelligence in the Dutch Press: An Analysis of Topics and Trends, Communication Studies. DOI: 10.1080/10510974.2020.1733038
Nisbet, Matthew C., (2008). Framing Science: A New Paradigm in Public Engagement. School of Communication, American University, Washington DC.
http://ion.uwinnipeg.ca/~clark/teach/zzArchives/3480/nisbetframingscience.pdf Retrieved: 2020-04-14
Obozintsev, Lucy. (2018). From Skynet to Siri: an exploration of the nature and effects of media coverage of artificial intelligence. University of Delaware, Department of
Communication.
http://udspace.udel.edu/bitstream/handle/19716/24048/Obozintsev_udel_0060M_13432.pdf?s
equence=1&fbclid=IwAR0Mc3Y8i6F1TnRk7D_i5U2Z23I2ugaOVZ0Qt-1FzUBnpUF0DNIW3IEqZjA. Retrieved: 2020-04-12
O'Leary, D. E. (2013). Artificial intelligence and big data. IEEE Intelligent Systems, 28(2), 96-99. https://ieeexplore.ieee.org/document/6547979. Retrieved: 2020-03-31
Schwab, Klaus. (2016). The Fourth Industrial Revolution: what it means, how to respond. The Economic Forum.
https://www.weforum.org/agenda/2016/01/the-fourth-industrial-revolution-what-it-means-and-how-to-respond/. Retrieved: 2020-04-01
Star, Susan Leigh. (1995) - Ecologies of Knowledge: Work and Politics in Science and Technology. Albany: State University of New York press. p 130.
Tortoise Media. (2020). The Global AI Index.
https://www.tortoisemedia.com/intelligence/ai/. Retrieved: 2020-04-12
Weissman, Mikael. (2015) Chinese Foreign Policy in a Foreign Perspective: A Responsible
Reformer “Striving for Achievement”. JCIR: VOL. 3, No. 1.
https://www.diva-portal.org/smash/get/diva2:817328/FULLTEXT01.pdf
Zhang, Baobao. Dafoe, Allan. (2019). Artificial Intelligence: American Attitudes and Trends. Center for the Governance of AI, Future of Humanity Institute, University of Oxford.
7. Appendix
Table A1: Views regarding artificial intelligence in American and Chinese newspaper (per
Table A2: Nisbet’s frames that Consistently Appear Across Policy Debates
Frame Defines Science-Related Issue As…
Social progress …improving quality of life, or solution to
problems. Alternative interpretation as harmony with nature instead of mastery, sustainability.”
Economic
development/competitiveness
…economic investment, market benefits or risks; local, national, or global competitiveness.
Morality/ethics …in terms of right or wrong; respecting or crossing limits, thresholds, or boundaries.
Scientific/technical uncertainty …a matter of expert understanding; what is known versus unknown; either invokes or undermines expert consensus, calls on the authority of “sound science,” falsifiability, or peer-review.
Pandoraís box / Frankensteinís monster / runaway science
…call for precaution in face of possible impacts or catastrophe. Out-of-control, a Frankenstein’s monster, or as fatalism, i.e. action is futile, path is chosen, no turning back.
Public accountability/governance …research in the public good or serving private interests; a matter of ownership, control, and/or patenting of research, or responsible use or abuse of science in decision-making, “politicization,”
Middle way/alternative path …around finding a possible compromise position, or a third way between conflicting/polarized views or options.
Conflict/strategy …as a game among elites; who’s ahead or behind in winning debate; battle of personalities; or groups; (usually journalist-driven
interpretation.)
Table A3: Article title and author from New York Times
Articles - the New York Times Month
1. The End Of Normal Michiko Kakutani
December 2019 2. When Robots Provide Care December 2019 3. A New Outlook on Antitrust Is
Taking Shape Steve Lohr
December 2019 4. While You Shop for Bargains, They
Will Be Hard at Work. By Michael Corkery, Sapna Maheshwari and Daniel Dorsa
5. Spot the Deepfake. (It's Getting Harder.) Cade Metz
November 2019 6. White-Collar Jobs Aren't Safe Either,
Jamie Condliffe
November 2019 7. Routes to New Insights on Race and
Identity, Lauretta Charlton
October 2019 8. Ready. Set. Write a Book. J. D.
Biersdorfer
October 2019 9. Why California Has Better Data
Protections. Natasha Singer
September 2019 10. White-Collar Jobs Were Supposed to
Go Overseas. They Didn't. Ben Casselman
September 2019
11. How Tech Can Make War Safer. LUCAS KUNCE
August 2019 12. Wanted: Spy. Employer: China. To
Apply: Click. EDWARD WONG
August 2019 13. He May Be Your Best Hope Against
the Killer Robots. CLAUDIA DREIFUS
July 2019
14. FaceApp: A Cautionary Tale. JOHN HERRMAN
July 2019 15. Buttigieg's Remarks on Police
Shooting Don't Satisfy Some Back Home. RIP GABRIEL
June 2019
16. U.S. vs. China: Why This Power
Struggle Is Different. Edward Wong June 2019 17. Google Relies On Underclass Of
Temp Labor. DAISUKE WAKABAYASHI
May 2019
18. Huawei Ban Gains Trump A Wall at Last. DAVID E. SANGER
May 2019 19. As Trade Talks Continue, China Is
Unlikely to Yield on Control of Data. Ana Swanson
April 2019
20. How to Take Back Control From Facebook. Charlie Warzel
April 2019 21. Countries Want to Ban
‘Weaponized’ Social Media. What Would That Look Like? Damien Cave
March 2019
22. Teaching a Generation of Machines, Far From the Spotlights of Silicon Valley. DAN BILEFSKY
March 2019
23. Playing Cat and Mouse With a Surveillance State. PAUL MOZUR
February 2019 24. A.I. Still Needs H.I. (Human
Intelligence). THOMAS L. FRIEDMAN
Note: All coding results from each article are available, contact author if desired).
Table A4: Article title and author from Forbes
Articles - Forbes Month
25. Connecting a Million Dots. William Baldwin
December 2019 26. At Forbes CMO Summit Europe,
Marketing Leaders Call For Peers
December 2019 27. To Take Risks To Drive Purpose.
Marty Swant, Forbes Staff
November 2019 28. Genius Drugs From Dumb Silicon.
Jillian D’Onfro
November 2019
29. The Forbes 400 October 2019
30. CLOTHING MARKET - WHEN THE ROBOTS DRESS US UP. Forbes
October 2019
31. HOT STUFF. Chloe Sorvino September 2019 32. The CrowdStrike Conspiracy: Here’s
Why Trump Keeps Referencing The Cybersecurity Firm. Rachel Sandler
September 2019
33. Personalizing Cancer. Noah Kirsch and Michela Tindera
August 2019 34. Shifting Shores. Yuwa
Hedrick-Wong
August 2019 35. Software Company Dynatrace Prices
Shares At $16 Ahead Of IPO. Kenrick Cai, Forbes Staff
July 2019
36. Jeffrey Epstein Wanted To Freeze Brain, Spread His DNA. Lisette Voytko,
July 2019
37. The New Nuclear. Christopher Helman
June 2019 38. Ag Sec. Sonny Perdue: We Want To
Be Most People-Focused Agency In The Federal Government. Kenrick Cai.
June 2019
39. IN A CLASH WITH THE GIANTS OF THE NETWORK.. From
Internet monitoring to quality chatbots. KRZYSZTOF DOMARADZKI
May 2019
40. EMOTIONS SELL, YOU HAVE TO UNDERSTAND THEM. Forbes.
May 2019 41. AND CLOSER THAN IT SEEMS April 2019 42. FROM PACKING. TO SERIOUS
BUSINESS - PACKHELP About six who reinvented the box
