Introduction
Artificial Intelligence (AI), a cornerstone of the Fourth Industrial Revolution, is a transformative force reshaping various fields, including education (Stone et al. 2022). Advances in machine learning and deep learning have altered traditional education practices. This trend accelerated during the Covid-19 pandemic, as online learning became a new normal, bringing the role of Artificial Intelligence in Education (AIEd), such as AI tutor and chatbots, into sharp focus. For example, OpenAI’s ChatGPT, praised for passing exams like the US Medical Licensing Exam (USMILE), demonstrates its educational potential. However, the misuse of it also sparked debates about academic integrity by blurring the line between assistance and plagiarism in paper writing (Bin-Nashwan et al. 2023). AIEd’s complexities make its representations in media important, as media discourses play a critical role in shaping public understanding and perceptions of technology (McCombs and Shaw 1972; Nelkin 1990).
While AIEd garners public interest, several critical gaps remain in its media representations. First, voices of social actors differ in their focus and do not carry equal weight, reflecting imbalances in power relationship (Coeckelbergh 2022; Foucault 2020; Knox 2020). Second, studies on AIEd media representations are scarce, since AI in education receives far less media coverage than AI in other fields. Third, current studies on AI media representations largely target “Western” nations, especially the US and the UK, further perpetuating geographical biases (Blanchard 2015; Brause et al. 2023).
To address these gaps, this study examines the representations of social actors in AIEd discourses in China Daily to probe the power dynamics and ideologies shaping public perceptions. China Daily was selected for this study due to its state-regulated media context, China’s leading position in AI, and China Daily’s role as a national-level English-language newspaper with a global reach (See Methodology for a more detailed rationale). These features ensure coverage that contrasts with predominantly Western-centric discourses and provides unique insights into Chinese perspectives on AIEd.
This study proposes the following research questions:
(1)
Who are the key social actors in AIEd coverage of China Daily, and how are they generally represented?
(2)
What discursive strategies are employed in the representation of social actors in AIEd discourses?
(3)
What power dynamics and ideologies are revealed through these representations in China Daily’s coverage of AIEd?
Literature review
Overview: technology, impact and media representation
Artificial Intelligence in Education (AIEd) incorporates AI technologies like machine learning (ML) and natural language processing (NLP) to personalize learning experiences and streamline administrative tasks (Brause et al. 2023). From early automated teaching machines in the 1960s to Intelligent Tutoring Systems (ITS) in the 1980s and 1990s, AIEd’s development advanced with the advent of ML and NLP in the twenty-first century (du Boulay et al. 2023). These breakthroughs enable adaptive learning platforms and virtual assistants, directly impacting students, teachers, and institutions (Holmes and Tuomi 2022). For instance, student-focused AIEd technologies, such as “content-based, exam-oriented” personalized AI learning machines, have gained popularity in China (Yuan 2024), while teacher- or institution-focused products remain in the pilot phase (Jing 2024).
The growth of AIEd in China is driven by government policies, edtech growth, and strategic collaborations (MEPRC 2018a, b; State Council 2017). Policies like the National Strategy for AI Development (NSAID) and the Education Informatization 2.0 Action Plan (EIAP2) promote cooperation among governmental, business, and educational stakeholders (Roberts et al. 2021). Global cloud service providers are enabling innovation and improvement in AIEd solutions developed by leading edtech companies like TAL and New Oriental (Williamson et al. 2023b). Policy shifts, such as ‘Double Reduction’ policy, push companies to adapt their products (Knox 2023). However, both top-down and market-driven mechanisms raise issues of equity and ethics (Gulson et al. 2022; Williamson et al. 2023b).
In this dynamic landscape, media representations shape perceptions of AIEd and its stakeholders (McCombs and Shaw 1972; Selwyn et al. 2020). Drawing on Foucault’s concept of discourse as embedded within power relations (Foucault 2005, 2013, 2020), AIEd discourses often bring together various voices, such as state-led visions and market-driven actions (Knox 2020). However, these voices differ in their focus and do not carry equal weight. Coeckelbergh (2022) cautions that AI development, often funded by governments and tech corporations, may leave citizens with little or no influence over the technology’s use and future, ultimately placing them in a state of unfreedom and despotism.
Despite growing interest in AIEd within social sciences, research on its media representation remains scarce, particularly concerning social actors and discursive strategies (Knox 2023; Williamson et al. 2023a). Existing studies often emphasize policy and commercialization but overlook how media frames these issues or represents diverse voices. Studies on AIEd discourse, like Nemorin et al. (2023), remain rare, and even this study focuses on public documents rather than media discourses of AIEd. This scarcity may stem from the gradual nature and perceived lower newsworthiness of education topics, which attract far less media attention compared to the immediate, headline-grabbing impact of AI breakthrough in other fields. As a result, limited media coverage further restricts the availability of research data. To address this scarcity, we have reviewed broader studies on AI media representations to provide context.
In the news media, AI often garners headlines through a lens of hype, that emphasizes its most sensational or controversial aspects (Elish and Boyd 2018; Lee et al. 2005). This can lead to a skewed public perception, sometimes amplifying AI’s current capabilities or potential risks. Previous research on AI media discourses in the U.S. and Europe shows a shift from initial optimism to a mixed perspective, weighing benefits against risks like ethical concerns and job replacement (Brennen et al. 2022; Chuan et al. 2019; Fast and Horvitz 2017; Zhai et al. 2020). In Asia, particularly China, media often presents AI in a positive way, focusing on local economic benefits rather than central directives (Cui and Wu 2021; Zeng 2021). However, Bareis and Katzenbach (2022) note that Chinese government sees AI as a key tool for maintaining social order through continuous monitoring, thereby reaffirming its paternalistic governance approach, enhancing governance efficiency and stability.
Current studies on AI media representations are heavily skewed toward Western contexts, leaving a significant gap in understanding how AI discourses are shaped in other cultural and regional settings. For example, Brause et al.’s (2023) analysis of over 600 articles on AI in media shows a disproportionate emphasis on North American (23%) and European (30%) perspectives, particularly from Western countries. This reflects an “ALIEN” bias—discourses rooted in American, Liberal, Industrialized, and Democratic societies—diminishing the visibility of AI’s impact in other regions (Blanchard 2015). Such geographical bias signals the need for a more diversified examination of AI discourses, particularly from distinct cultural and national contexts to achieve a balanced understanding.
To address these gaps, this study examines the representations of social actors in AIEd discourses in China Daily to uncover the power relations and ideologies shaping public perceptions. By investigating the discursive framing of AIEd—whose voice is selected and who speaks for or against it—can unveil the power and marginalization dynamics that shape educational policy and public opinion.
Newspaper discourse: the shaper of public opinion
Newspapers have the power to define this technology, assign value to it, and shape public opinion regarding its use and implementation (Fairclough 1995; Richardson 2013). Media shape public opinion by influencing what the public considers important and how they think about issues (Lippmann 2017). Theories like cultivation theory (Gerbner 1969), third-person effect (Tal-Or et al. 2009), and agenda setting (McCombs and Shaw 2005) support this view, emphasizing that news sources and construction are closely linked to the actions and opinions of powerful social groups (Richardson 2017), thereby shaping public’s perceptions of reality.
In China, the media’s role has evolved significantly, from an ideological weapon to a market commodity, and now a conduit to go global (Sen 2021). Drawing on Foucault’s (2005) concept of power as pervasive and exercised through discourses, newspaper legitimizes certain perspectives while sidelining others (Entman 1993; Van Gorp 2007). These legitimized perspectives often align with dominant group, which termed by Gramsci (2020) as “hegemony”. In Fairclough’s framework, this constitutes “discursive hegemony” (Fairclough 2003), where media actively normalizes technological agendas, commercial interests, or policy directives (Williamson et al. 2023a).
From this perspective, the media are not only observers but also participants in constructing discourses. By amplifying or neglecting voices of certain social actors, media can reproduce structural inequalities, which limit alternative path to new models of education and innovation (Van Dijk 1995). This power-marginalization lens is key to understanding which actors gain visibility in AIEd discourses and how their roles are portrayed. Thus, media representations of AIEd, therefore, are never unproblematic but situated within broader power relations that influence public perceptions and policy directions.
Intertextuality and social actor theory
Critical discourse analysis (CDA) tools examine both explicit power in discourse (e.g., who is quoted as an authority) and the covert power behind discourse (e.g., who decides what is “newsworthy”), revealing how inequalities are reproduced or contested (Wodak and Meyer 2015).
Intertextuality in critical discourse analysis
Intertextuality, rooted in the works of Bakhtin, Kristeva, and Barthes, suggests that texts gain meaning only in relation to other texts and social practices (Allen 2011; Fairclough 1995). Building on this foundation, Chouliaraki and Fairclough (2021) illustrate how intertextuality produces heterogeneity in texts through the coexistence of multiple voices and genres. Richardson (2017) further differentiates between external intertextuality, concerning texts within larger social practices, and internal intertextuality, addressing how quotations, reported speech and references appear within a text. Gadavanij (2002) applied Fairclough’s categorization of internal intertextuality (mixed, sequential, and embedded) to Thai no-confidence debates and argue that these forms serve as discursive strategy to fulfill specific communicative functions. Similarly, Gong et al. (2023) employs intertextuality to analyze how multiple perspectives are endorsed, contested or aligned within a single discourse.
This study applies intertextuality analysis, particularly internal intertextuality, to track representations of marginal voices, their alignment or conflict with dominant discourses, and the co-occurrence relationships between social actors. Here, mixed intertextuality refers to a blending of voices within a single sentence. Sequential intertextuality explores how perspectives are arranged across stages of a text, while embedded intertextuality examines how external voices are incorporated into institutional discourses. These categories are adapted to analyze how AIEd discourses are aligned, integrated, and embedded in the media discourse, revealing power relationships and dominant discourses in AIEd representations.
Van Leeuwen’s social actor theory
Van Leeuwen’s social actor theory complements intertextuality analysis by providing a systematic framework to examine how key social actors are represented in discourse (Van Leeuwen 2013). It provides a taxonomy for inclusion strategies (e.g., activation vs. passivation, personalization vs. impersonalization) and exclusion strategies (e.g., suppression, backgrounding), demonstrating which actors are foregrounded or marginalized. For example, Gong et al. (2023) analyzed how China Daily represented social actors in its Covid-19 coverage, while Sunday and Fagunleka (2024) explored representations of social actors in economic and financial crime reports in Nigerian newspapers. Figure 1 visualizes the system network of these strategies and their application in media discourse.
Fig. 1: The representation of social actors in discourse: system network.
figure 1
This figure shows Van Leeuwen’s system network for representing social actors in discourse, categorizing strategies into inclusion, exclusion, personalization, and impersonalization, which are further divided into detailed subcategories.
Full size image
Integrating intertextuality and social actor theory
While van Leeuwen’s social actor theory focused on structural representations, it does not fully address how social actors’ voices are quoted or recontextualized in media discourses. To bridge this gap, this study integrates intertextuality analysis with social actor theory to analyze marginalized voices and their co-occurrence patterns, as well as prominent social actors such as government and corporate. This integration uncovers both the textual relationships (through references and quotations) and representational strategies (through activation, passivation, etc.), illuminating how power relations and marginalization are constructed in China Daily’s AIEd coverage.
Methodology
Data collection
Data for this study were collected from the Nexis Uni database, targeting articles related to Artificial Intelligence in Education (AIEd) published in China Daily from June 1, 2018, to December 1, 2023. A total of 100 articles were identified, resulting in a corpus of 57,456 tokens, covering significant developments in AIEd, such as the advent of GPT models and the impact of the Covid-19 pandemic.
The choice of China Daily is justified for three main reasons. First, Chinese media operates under state-regulation model, which contrasts with the commercially driven discourses of Western media (Jeffrey and Sen 2020). Second, China is a leading nation in AI development, offering perspectives beyond predominantly Western-centric discourse. While China Daily does not represent all of China’s viewpoints, it offers a valuable lens into key aspects of China’s stance on AIEd, thus complementing and balancing Western perspectives. Third, as China’s sole national-level English-language daily newspaper with global circulation, China Daily is uniquely positioned to present China’s perspectives to global audiences. This makes it an ideal subject for studying AIEd discourses and how social actors are portrayed in a context distinct from Anglophone media, contributing to more inclusive insights regarding AIEd discourses.
June 1, 2018, was selected as the starting date for this study based on an analysis of China Daily coverage trends retrieved from Nexis Uni using relevant keywords. The analysis showed a noticeable increase in AIEd-related reporting from mid-2018, indicating growing interest in integrating AI into educational practice and policies. This surge coincided with key advancements in AI technology, such as the release of OpenAI’s GPT-1, which also spurred greater focus on developing home-grown AIEd solutions in China to address local educational needs and support national modernization goals.
The dataset was constructed using targeted search terms based on the AIEd taxonomy by Holmes and Tuomi (2022). These search terms included combinations of terms related to “Artificial Intelligence” and “Education”, ensuring a broad and relevant selection of AIEd applications. Search parameters focused on newspaper articles in the “Education & Training” category, excluding irrelevant content. Only English-language publications were included to maintain consistency.
To ensure balanced representation, the study period was divided into yearly intervals. A proportional stratified sampling method was applied, selecting 25% of the deduplicated articles from each year according to the relevance ranking provided by Nexis Uni, resulting in a total of 100 articles. This approach was chosen to avoid overrepresentation of any single year and capture longitudinal trends. After applying a deduplication process in Nexis Uni, a manual review removed minor residual duplicates to ensure dataset accuracy.
Data analysis procedure
This study applies a mixed-method approach—combining identification of social actor voice, intertextuality analysis, and social actor analysis—to investigate AIEd representation in China Daily within a broader discursive and socio-cultural context. Voices identification and intertextuality analysis address RQ1 by examining how social actors are represented, while social actor analysis addresses RQ2 by exploring the discursive strategies employed. Situating these methods within the socio-cultural context addresses RQ3, uncovering how various social actors are portrayed, by whom, and with what power and ideological implications.
Identification of social actor voices
Voices identification was conducted through displaCy’s Named Entity Recognition (NER) tool, which processes the articles to identify and categorize social actors based on entities such as person names, organizations, and locations. Given the potential ambiguity in recognizing smaller companies or less frequently mentioned individuals, manual verification was concurrently conducted during the initial processing with displaCy to ensure accuracy, following pre-established criteria such as entity relevance, contextual accuracy, and coding consistency. To further enhance reliability, 30% of the data was independently reviewed, achieving over 90% intercoder reliability.
Discursive strategies analysis
After social actor voices are identified, intertextuality analysis and social actor analysis are applied to understand how social actors are presented by identifying discursive strategies employed.
Intertextuality analysis focused on the direct and indirect reported speech of social actors, identifying connections and patterns through mixed, sequential, and embedded intertextuality. This approach examined marginal voices, their co-occurrence patterns, and the linguistic strategies used to frame their speech, revealing how different texts reference and respond to social actors, contributing to their portrayal in AIEd discourses.
Social actor analysis, however, focused on categorizing the representation of key social actors, such as government, corporate, and educational institutions. Using van Leeuwen’s Social Actor Theory, this analysis explored how each actor is portrayed through discursive strategies such as activation, passivation, and other forms of representations. Nvivo was employed to extract and code relevant sentences associated with key social actors (institutions and individuals, including abbreviations and pronouns).
Tables 1 and 2 present the taxonomy for analyzing social actor representations, based on Van Leeuwen’s (2013) social actor theory. Building on Fig. 1 “The representation of social actors in discourse: system network”, these tables outline the subcategories of inclusion, such as activation, passivation, personalization, and impersonalization, which serve as key criteria for data coding and analysis in this study.
Table 1 Taxonomy of social actor representation.
Full size table
Table 2 Subcategories of inclusion.
Full size table
Activation and passivation indicate how social actors are framed—as agents of change (e.g., government issuing AIEd policies) or as recipients or passive beneficiaries (e.g., students and parents). This distinction reflects how agency is distributed, with institutions often depicted as powerful decision-makers while individuals are marginalized.
Similarly, personalization versus impersonalization reveals whether actors are humanized (e.g., Mi Wenjuan, CEO of VIPKid) or abstracted into categories (e.g., “major tech giants”), shaping readers’ emotional engagement and perceptions of responsibility.
Results
The results of this study are classified into three sections, each addressing one of the research questions: (1) the key social actors and their general representations in China Daily (RQ1); (2) the discursive strategies employed in the representations of social actors in AIEd discourses (RQ2); and (3) the power dynamics and ideologies underlying these representations (RQ3).
Key social actors and their representation in China Daily
This section seeks to address RQ1: Who are the key social actors in AIEd coverage of China Daily, and how are they generally represented?
Dominant voices: corporate and government actors
Results show that the AIEd discourses in China Daily are dominated by corporate and governmental actors.
Of 100 articles analyzed, individual social actors quoted in China Daily’s AIEd news reports were coded and categorized into five groups: company executives (86 actors), Chinese officials (43 actors), education administrators (26 actors), researchers and professors (31 actors), and the public (24 actors). Together, company executives and Chinese officials made up more than 60% of all quoted actors. Individual members of the public contributed only 11% of the total (for detailed proportions, see Fig. 2).
Fig. 2: Proportion of quoted social actors in AIEd discourse.
figure 2
This pie chart displays the distribution of quoted social actors in AIEd discourse in China Daily, categorized into five groups: corporate executives, government officials, education administrators, researchers, and the public. Corporate executives are the most frequently cited group.
Full size image
In addition to individual actors, corporate and government institutions also dominate AIEd discourses, with 108 companies mentioned 450 times, and 47 government entities mentioned 99 times. This reflects the institutional alignment driving AIEd discourses, reinforcing the dominance of corporate and governmental perspectives.
The analysis ranked the top five corporate actors whose name appeared most frequently in China Daily AIEd discourses: Tencent, TAL Education Group, VIPKid, Zuoyebang, and SenseTime (five leading EdTech companies in China) (refer to Fig. 3). Companies frequently mentioned in coverage before 2021 were TAL and VIPKid, but post-2021 discourses shifted focus on AI-driven companies like SenseTime, indicating a shift in EdTech landscape.
figure 3
This bar chart shows the frequency of mentions of key corporate and governmental actors in AIEd discussions, with the Ministry of Education (MOE) and Tencent being the most frequently cited entities.
Full size image
Table 3 presents the roles and engagement of quoted executives in the top five frequently mentioned EdTech companies, such as Zhang Bangxin (TAL), and Mi Wenjuan (VIPKid) during the early coverage. Analyzing the discursive strategies with both institutional and individual actors, as well as their shift from presence to relative absence in the coverages, provide insights into deconstructing power in AIEd discourses.
Table 3 Corporate social actors (CSA) mentioned in AIEd discourse.
Full size table
Governmental actors, such as Ministry of Education (MOE) and the State Council, also played a central role. Twenty articles mentioned the MOE (surpassing even Tencent), and officials like Huai Jinpeng (MOE) were frequently quoted. Table 4 provides the quoted government figures in the top five frequently mentioned government institutions, and their titles as complements to these institutional names. Such specific identification can serve as a basis for analyzing patterns of discursive strategies that can elucidate certain connections of dominance and influence.
Table 4 Government social actors (GSA) mentioned in AIEd discourse.
Full size table
Marginalized voices and co-occurrence patterns
Public voices, while present, are underrepresented, comprising only 11% of quoted social actors. Parents’ voices dominate this category, with minimal representations from students.
Co-occurrence analysis shows the extent to which social actors are often positioned alongside other social actors, constructing relationships between them. Prior to 2021, corporate figures frequently appeared alongside public voices, especially parents, aligning corporate discourses with parental perspectives. This pattern was particularly prominent before 2021 but declined (Fig. 4). This shift suggests a reconfiguration of discourse, reflecting broader socio-political shifts, including the “Double Reduction” policy, which reshaped the EdTech landscape and patterns between corporate and public voices.
figure 4
This figure compares the co-occurrences of different social actors in AIEd discourse before and after 2021, highlighting a notable decline in corporate-public interactions after 2021.
Full size image
Public voices, while occasionally used to support corporate discourses, lack diverse perspectives and are largely absent in critical discussions. Their marginalization reflects a top-down approach to discourse, where corporate and government actors shape public understanding, leaving little room for alternative visions of AIEd.
Discursive strategies for representing social actors in AIEd discourse
This section addresses RQ2 by examining how the voices of social actors are discursively constructed. It begins with an intertextuality analysis to explore the representation of marginal voices and co-occurrence patterns to uncover alignment or tension between different social actors. Then, Van Leeuwen’s social actor theory is employed to analyze how government and corporate actors are portrayed, focusing specifically on inclusion strategies, such as activation, passivation, personalization, and impersonalization, as outlined in Section “Data analysis procedure” (see Tables 1 and 2).
Intertextuality analysis of marginal voices and their co-occurrences
News often incorporates quotations from sources. However, not everybody’s voice gets published in the news. Media has its selective role in allowing some people’s access to communicate with the public while denying others (Fairclough 1995; Richardson 2017). Intertextuality analysis illuminates how different voices are aligned, obscured, or integrated into dominant discourses (Gong et al. 2023; Li 2009; Teo 2000). Mixed intertextuality reconciles different stakeholders while obscuring conflicts. Sequential intertextuality steers readers to dominant discourses, and embedded intertextuality incorporates public voices into institutional narratives, reinforcing dominant ideologies.
To further understand the co-occurrence between public voices and corporate executives, each news article featuring these social actors was examined. Table 5 provides an overview of the intertextuality types observed in the selected excerpts.
**Table 5 Intertextuality analysis of social actor representation in
China Daily
.**
Full size table
Excerpt 1 (Intertextuality)
“‘One important reason that we choose to take extra classes online is that our child can get one-on-one tutoring without wasting time in traffic, especially in Beijing,’ said Wang Liping, a mother of a 15-year-old boy from Beijing.” (“Online education gaining momentum,” August 21, 2018)
Excerpt 2 (Intertextuality)
“‘The company is striving to advance education through technology, hoping to build a large-scale, low-cost and high-quality online education experience,’ said Zhang Bangxin, founder and CEO of TAL.” (“Online education gaining momentum,” August 21, 2018)
Excerpts 1 and 2 are from the same article, discussing China’s rapidly growing online education market, which results from high parental expectations for students’ competency in the academic race. These excerpts demonstrate internal intertextuality, where corporate and parental voices are intertwined within the same article to align parental needs with corporate goals.
In Excerpt 1, Wang Liping’s statement exemplifies mixed intertextuality by integrating personal concerns about time efficiency with a larger societal problem (i.e., traffic congestion in Beijing). Her casual tone and everyday language (“without wasting time”) establish credibility, while first-person plural pronouns (“we,” “our”) construct a collective parental identity, suggesting shared concerns.
In Excerpt 2 shifts to corporate voice, with Zhang Bangxin employs embedded intertextuality to reframe these parental concerns within TAL’s corporate mission. Formal and institutional language (“large-scale, low-cost and high-quality”) projects the company’s objectives as both progressive and pragmatic, contrasting with the personal tone in Excerpt 1. This shift from personal to institutional voice exemplifies sequential intertextuality, linking individual needs to corporate strategies.
Together, these excerpts demonstrate how differing registers—Wang’s relatable, conversational tone and Zhang’s corporate, formal style—work in conjunction to legitimate corporate ends through the incorporation of parental concerns. This interplay of intertextual voices locates personal experiences within broader corporate discourses, jointly constructing a cohesive story about online education as a remedy for societal pains.
Excerpt 3 (Intertextuality)
“For me, building a toy robot is more meaningful than playing with a toy car. A kid can thus learn to develop his or her logical thinking and design skills while playing.” (Qin Liu, a mother of 6-year-old boy) (“Brawny robots for brainy tots,” January 14, 2019)
Excerpt 4 (Intertextuality)
“‘Our business is to combine technology with education to teach kids how to build robots and how to code,” said Wang.” (Wang Jianjun, the CEO of Makeblock) (“Brawny robots for brainy tots,” January 14, 2019)
Excerpts 3 and 4 are from the same article, discussing how Chinese parents favor early education in coding and robotics, with companies like Makeblock providing tools to develop children’s skills through interactive learning.
In Excerpt 3, Qin Liu’s statement also demonstrates mixed intertextuality by weaving parental appreciation of robots as educational tools with broader developmental goals for children. Her colloquial tone (“for me,” “while playing”) and descriptive verbs (“develop,” “learn”) frames robot building as both practical and aspirational. The first-person narrative reinforces the parent’s voice as an authentic endorsement, situating the discourse in individual experience.
In Excerpt 4, Wang Jianjun’s statement embodies embedded intertextuality, incorporating the parental perspective into the corporate storyline. The phrasing “our business is to combine technology with education” formalizes the parent’s appreciation in Excerpt 3 as part of Makeblock’s strategic vision. Evaluative verbs such as “teach” and “combine” place the company as an innovator responding to a societal demand for educational tools. The absence of personal pronouns and the appeal to collective aspirations via “kids” and “how to code” reinforces the company’s institutional authority. Together, Excerpts 3 and 4 demonstrate how internal intertextuality aligns parental appreciation with corporate goals, normalizing educational technology as a societal solution.
Comparing Excerpts 1 and 2 with Excerpts 3 and 4 manifests external intertextuality, as similar themes of alignment and institutional authority recur across articles. These strategies collectively reinforce dominant discourses, embedding individual needs within corporate-driven solutions.
Representation of key social actors
This section examines how key social actors are systematically represented through impersonalization, functionalization, activation, and passivation. The excerpts illustrate different discursive strategies used to portray government and corporate actors. As shown in Table 6, corporate (98%) and government actors (97%) are mainly represented through activation, highlighting their roles as proactive agents. Impersonalization is also predominant, with corporate (74%) and government actors (62%) portrayed in institutional identity, emphasizing their group authority.
**Table 6 Representation of prominent social actors in
China Daily
.**
Full size table
Impersonalization
Social actors are often impersonalized by stripping away human qualities, reducing them to abstract concepts or objects, particularly in official statements or when the source of information is being concealed (van Leeuwen 2008). Impersonalization can be achieved through abstraction, representing actors by assigned qualities, or objectivation, linking them to associated places or things. In this study, utterance autonomization—representing actors by their statements—is the primary form of objectivation analyzed. This method distances the social actor from the action, emphasizing formal authority nature, reducing accountability and enhancing perceived impartiality. Impersonalization and objectivation also align with institutional power structures, framing actions or decisions as systemic and neutralizing emotional engagement.
Excerpt 5 (Objectivation):
“The State Council has announced measures to encourage the growth of online education in a major boost to businesses in the sector, with experts calling for administration to be streamlined to further invigorate the market and promote education equality.” (“Cabinet boost for only education,” September 3, 2019)
Excerpt 6 (Objectivation-Utterance autonomization):
“The plan stated that China will launch major research projects in artificial intelligence, quantum information technology, integrated circuits, life science and health, aerospace technology, and deep earth and ocean exploration, as well as other cutting-edge fields.” (“Future technology schools’ get green light,” June 5, 2021)
Excerpts 5 and 6 are from articles that stress China’s strategic use of technology for national development, the State Council advancing online education and the Five-Year Plan setting priorities for cutting-edge research. Both excerpts illustrate how governmental entities are objectivated. The “State Council” is objectivated through its institutional authority, emphasizing the official and formal nature of the initiatives while backgrounding individual human actors. The phrase “The plan” employs utterance autonomization, objectifying the intentions and strategies of policymakers who drafted China’s 14th Five-Year Plan.
Functionalization
Functionalization represents social actors based on their occupation or role, realized through noun formations, place or instrument associations, and generalized categories (van Leeuwen 2008).
Excerpt 7 (Functionalization)
“As a major tech giant in the industry, Tencent has strong cloud computing technologies and the ability to inspire innovation in the market.” (“Tencent launches training school to nurture AI talents,” September 20, 2019)
Excerpt 8 (Functionalization)
“To help solve such problems, Chinese AI pioneer SenseTime is also offering a string of free online AI curriculum options for students and teachers amid the COVID-19 pandemic.” (“AI-enabled curriculum helps nurture young talent,” August 5, 2020)
Excerpts 7 and 8 demonstrate how firms such as Tencent and SenseTime are functionalized, in this case with respect to their contributions to ‘AI education.’ Tencent is functionalized as a “major tech giant” following the debut of Tenshi School, which aims to nurture AI talents. This functionalization emphasized Tencent’s role in leveraging its technological strength to address China’s AI talent gap. In Excerpt 8, SenseTime is functionalized as the “Chinese AI pioneer,” with detailed evidence of its work with schools to provide free online AI curriculum options during the pandemic.
Activation
Activation highlights the active roles of government and corporate actors in advancing AIEd, as seen in Excerpts 9, 10, and 11, where the Ministry of Education and tech giants like Alibaba and Tencent are presented as key players in advancing AIEd.
Excerpt 9 (Activation)
“In April, the Ministry of Education issued an action plan to promote AI education at universities.” (“Online education gaining momentum,” August 21, 2018)
Excerpt 10 (Activation)
“The ministry said 269,911 after-school institutions, or 98.9 percent of the 272,842 establishments that were found to be problematic, had completed rectifications by Dec 30.” (“Oversight tightened on expat staff at cram schools,” July 16, 2019)
Excerpt 11 (Activation)
“Representatives from internet giants including Alibaba, Baidu and Tencent also attended, sharing their explorations in empowering children with frontier technologies.” (“Forum looks at opportunities and challenges facing preschool sector,” December 10, 2022)
The context for Excerpts 9,10, and 11 includes the rapid expansion of China’s online education market, which is projected to exceed $104 billion by 2025, the tightening of regulations of foreign teachers at after-school training institutions, and advancements in China’s preschool education.
Activation is realized through participation in grammatical participant roles, where the active role of the social actor is prominently foregrounded. This can be realized by transitivity structures or realized through prepositional circumstantial, such as “by” or “from.” In Excerpt 9, the Ministry of Education is activated as the Actor in the material process “issued,” emphasizing its active involvement in advancing AI education through policy implementation. In Excerpt 10, the Ministry of Education is activated as the Sayer in the verbal process “said,” showing its active role in reporting the completion of rectifications by a majority of after-school institutions. In Excerpt 11, representatives from Alibaba, Baidu, and Tencent are activated through the preposition “from,” emphasizing their participation in sharing technological advancements. This stresses the collaborative effort between the government and tech companies to enhance preschool education with innovative solutions, reflecting a joint initiative to support children’s development.
Passivation
Passivation, on the other hand, often depicted social actors as recipients of actions, either subjected or beneficialized. Subjected social actors are presented as objects in the representation, while beneficialized social actors are third parties who benefit positively or negatively from an activity.
Excerpt 12 (Subjection)
“While it marked the world’s largest-ever fundraising in the online education sector at that time, the firm (VIPKid) was said to have encountered difficulties in raising new funds after then.” (“VIPKid gets new funding led by Tencent,” October 9, 2019)
Excerpt 13 (Beneficialization)
“The fresh fundraising pushed the company’s (VIPkid) valuation over 20 billion yuan, making it the biggest online education company globally.” (“Online education gaining momentum,” August 21, 2018)
Excerpts 12 and 13 show how companies like VIPKid are passivated. In Excerpt 12, VIPKid is subjected through the phrase “was said to have encountered difficulties,” which passivates the firm by presenting it as experiencing challenges imposed by external circumstances, rather than as an active agent. In Excerpt 13, VIPKid is beneficialized as the Recipient in the material process “pushed the company’s valuation,” which positively impacts its status by establishing it as the largest global online education company.
Power dynamics and ideologies in AIEd discourses
This section addresses RQ3 by exploring the power dynamics and ideological implications associated with the representations of social actors in China Daily’s AIEd coverage. Based on findings in Sections “Key social actors and their representation in China Daily” and “Discursive strategies for representing social actors in AIEd discourse”, we examine how government, corporate and public voices interact to construct dominant discourses while marginalizing alternative perspectives.
Hierarchical collaboration: state-driven corporate roles
The way how government and corporate actors are represented in China Daily reveals a hierarchical collaboration, where state policies guide the direction of AIEd development and establish the framework within which corporate actors can innovate, attract investment, and implement their solutions.
Governmental actors are consistently portrayed as authoritative and proactive, often activated to emphasize their leadership. For example, in 2019, “The State Council announced measures to encourage the growth of online education” (Excerpt 5), demonstrating its leadership in steering the AIEd landscape.
This top-down model is reflected in the way corporate actors align their strategies with state-led aspirations, utilizing the regulatory environment and policy alignment to secure market opportunities and investor confidence. Companies like VIPKid achieved significant milestones during this period, with “fresh fundraising pushing VIPKid’s valuation over 20 billion yuan, making it the biggest online education company globally (Excerpt 13).
Corporate actors, while functionalized as “industry pioneers” (Excerpt 7 and 8), are often contextualized within governmental frameworks, suggesting a cooperative but subordinate role in advancing national objectives. The introduction of the 2021 “Double Reduction Policy” marked a pivotal shift, redirecting corporate priorities from profit-driven to state-alignment first. This policy not only restricted education commercialization but also highlighted government authority. When MOE was cited that “98.9% of after-school institutions were problematic” (Excerpt 10), companies that failed to adjust their product to comply with state regulations “encountered difficulties in raising new funds” (Excerpt 12) and faded out of the coverage of China Daily. In contrast, those that adhered to government objectives, such as SenseTime, were consistently portrayed in a positive light.
Marginalized public voices in AIEd discourse
Public voices, particularly those of parents and students, remain underrepresented, accounting for only 11% of quoted actors in AIEd discourses (Fig. 2, Section “Key social actors and their representation in China Daily”). When mentioned, parental perspectives are often embedded within discourses that serve corporate or state interests. Before 2021, parents were framed as passive beneficiaries of EdTech innovations, aligning their needs with corporate goals. However, this alignment declined sharply after the implementation of the “Double Reduction Policy”, and the shift in media focus further sidelined parental concerns, emphasizing state-driven objectives such as educational equity over corporate solutions.
Students, particularly those of K-12 age, face even greater marginalization. Their voices are largely absent, and only a few undergraduates appear as symbols of future potential. For example, students are often represented as beneficiaries of early exposure to AIEd, aligning with the state’s vision for cultivating long-term talent. The focus on their future potential diverts attention from their current needs, neglecting critical discussions about equity and inclusivity in AIEd implementations. Moreover, such representations avoid ethical concerns, such as the impacts of AIEd on student agency, well-being and their day-to-day classroom experiences.
The representations of dominant and marginalized actors in China Daily’s AIEd discourses reflect underlying ideological frameworks. Government actors are constantly mobilized to emphasize state control, framing educational innovation as a state-led endeavor. This discourse reinforces the government’s authority to regulate, shape, prioritize AIEd development. Corporate actors, in this context, become instruments of state goals and enablers to state-endorsed ends, with their market autonomy constrained by the need to align with national interests and government-directed reforms. In this hierarchy structure, public voices, especially those of parents and students, are largely sidelined. Such representations diminish local agency or alternative views, prioritizing hierarchy and institutional mechanisms as the primary framework, which lends itself to a systemic bias.
Discussion
This discussion situates the textual finding within socio-cultural context, examining how dominant social actors, such as corporations and governments, collaborate hierarchically in the context of AI in education (AIEd), while marginalizing public voices. It reflects underlying power structure and ideological framing.
State leadership and corporate’s adaptive role in AIEd discourses
In contrast to Cui and Wu’s (2021) findings that Chinese media’s AI discourse is shaped more by local economic factors than centralized government directives, this study suggests a hierarchical collaboration between government and corporate actors, highlighting the state’s leadership and corporate’s adaptive role. This finding, however, is consistent with Du’s (2018) model of governance, which describes the interaction between the “invisible hand” of market forces and the “visible hand” of government intervention.
Government actors in China Daily are portrayed as authoritative drivers of education equity and technological advancement, which is partially in line with Bareis and Katzenbach’s (2022) findings on Chinese government’s use of AI to maintain social order, although with different focus. For example, state’s leading role is evident in Ministry of Education’s implementation of “double reduction” policy, which redirects the commodification of education, and reinforces the ideology of state-led modernization (Knox 2021).
The pursuit of educational equity is achieved through commercial endeavors, aligning with Zhu et al.’s (2024) concept of state entrepreneurship, where national objectives are realized through market-enabled actions. Corporate actors in China Daily are functionalized as “tech giants” and “industry pioneers,” aligning their innovation with state-led objectives and contributing to national goals. Their visibility in media discourse is shaped by compliance with government policies, reflecting their subordinate position.
For example, SenseTime’s free AI courses embody the aspiration outlined in policies like Education Informatization 2.0 Action Plan and the broader goal of popularizing education resources under the “double reduction” policy. SenseTime’s compliance ensures its consistent positive portrayal as adaptable and relevant. Studies have identified it as China’s most valuable unicorn and the world’s top AI start-up by value (Knox 2023; Ko 2019). In contrast, VIPKid, once a prominent actor before 2021, have seen a sharp decline in visibility in China Daily due to their inability to adapt to new regulations, revealing the consequences of misalignment with state polices.
This shift in media focus reflects the state’s effort to curb education commodification and promote educational equity. It embodies the Foucauldian notion of “normalized power”, where the state expands its regulatory reach by constraining private firms. This mirrors Knox’s (2023) findings on Chinese government’s recalibration of edtech toward non-profit operations, reducing the dominance of commercial motives. While these measures reduce financial burdens for many parents, critics argue they may limit affordable tutoring options for some lower-income families, showing how well-intentioned centralized policies can produce uneven local impacts.
Marginalization of public voices in AIEd discourses
The results show that public voices (e.g., parents and students) are notably underrepresented in China Daily’s AIEd discourses, which highlights a lack of diverse perspectives and potential biases in the media portrayal of AIEd. Parents’ voices emerge sporadically but typically co-occur with corporate discourses that frame them as consumers whose needs can be by EdTech products, failing to emphasize their concern or agency. Students, in comparison, are largely absent from these discourses. When mentioned, students are depicted as future beneficiaries of AIEd, aligning with the state’s vision for long-term talent cultivation.
Parents serve a dual role in China Daily. On the one hand, they are portrayed as active adopters of EdTech, emphasizing practical benefits brought by online tutoring, AI-assisted tools, and coding programs. These discourses reflect their reliance on such technologies and unintentionally endorse the legitimacy of EdTech commercialization. On the other hand, parental anxiety about education, such as the pressure of “neijuan” (academic rat race), is linked to their support for government interventions like the “double reduction” policy. This dual framing positions parents as both beneficiaries of AIEd products and endorsers of state-led policy shifts, echoing Coeckelbergh’s (2022) notion of “unfreedom”, where citizens’ influence over the technology is constrained by powerful government and corporate discourses.
While China Daily emphasizes efficiency and state-led equity, it overlooks issues like cost or accessibility. This reflects Knox’s (2021) critique of technical nationalism, which refers to the prioritization of national technological advancement over addressing broader social issues. For example, the double reduction policy, though well-intentioned and effective in curtailing private tutoring, reduces affordable shadow education by driving it underground as illegitimate one-on-one tutoring arranged through personal networks, which are less accessible to disadvantaged groups. This framing risks obscuring systemic challenges, such as the exclusion of rural or marginalized communities, reflecting Van Dijk’s (1998) theory on ideological framing, where dominant discourses overshadow public concerns, normalizing the commodification of education and limiting the scope of grassroots agency. Moreover, China Daily largely skips the topic of financial burdens on families, despite evidence that Chinese parents invest heavily in education. For example, Dönmez (2024) found that Chinese parents spend approximately 20% of household income on K-12 education, including AIEd, as part of national modernization goals.
Student voices in China Daily are relatively absent, particularly those in the K-12 sector. When they do appear, it is often as university graduates who benefited from early exposure to AIEd, aligning with the state’s long-term talent development goals. This limited portrayal contrasts sharply with the investor-driven perspective described by Williamson et al. (2023a), where students are commodified as data sources or digital assets within broader EdTech frameworks. Unlike The New York Times, which amplifies student voices through platforms like the Student Opinion Forum, China Daily lacks similar channel, further marginalizing students’ perspectives in its AIEd discourse.
Conclusion
This study examines the representation of social actors in AIEd discourses in China Daily (2018–2023) using van Leeuwen’s Social Actor Theory and intertextuality. The findings address three research questions.
For RQ1, the results show that corporate and government actors dominate the discourse while public actors are underrepresented. The co-occurrence pattern between parental needs and corporate offerings reflects an initial education commercialization, where parents are framed as consumers and students as future beneficiaries of EdTech products.
For RQ2, intertextuality analysis reveals how commodification of education was normalized through the alignment of parental and corporate voices. Discursive strategies, such as the activation of government actors as leaders (e.g., the Ministry of Education issuing action plans) and the functionalization of corporate actors as contributors to state-led goals, construct a collaborative partnership. Public actors, by contrast, are often depicted as beneficiaries, with limited agency in shaping AIEd discourses.
For RQ 3, the study on power dynamics and ideologies uncovers a hierarchical structure, where China Daily represents state’s leadership in shaping AIEd discourses and corporate adaptive role in aligning their innovation with state-led objectives Policies serve not only as mediators among stakeholders, but also as tools for steering ideological directions, emphasizing educational equity and national rejuvenation. However, the marginalization of public voices, including their agency and immediate concerns, risks sidelining alternative perspectives and reinforcing a systemic bias that prioritizes institutional mechanisms over grassroots concerns.
This study contributes to our understanding of AIEd media discourses, a underexplored field. It also addresses the predominantly Western-centric focus found in existing studies on AI media representations. These findings unpack the power relations and ideologies that underpins AIEd discourses, reflecting the motivation, aspiration, and concerns of different social actors. Honoring these diverse voices can help to create an equitable and inclusive AIEd development.
Beyond academic contributions, policymakers, educators, and media professionals can draw on this study’s insights to advocate a more inclusive and sustainable discourse around AIEd, one that respects the regional contexts and addresses the unique needs of different regions.
This study, however, has limitations. Its exclusive focus on China Daily, the only national English-language newspaper as a conduit for China’s AIEd discourses to international audiences, risks overlooking localized or cultural-specific discourses embedded in regional or local-language media. Future work could extend this beyond the English and corporate media to include local-language or social platform outlets and a broader range of perspectives and audience reception. As a basis for comparative studies, this study invites cross-cultural analyses to explore how cultural and political contexts shape AIEd discourses globally.