Abstract
With the promotion of AI applications, people have undergone various changes, especially high school students who have a strong sense of engagement in AI-related activities. In order to explore the interactive mechanism and influencing factors between AI literacy and computing thinking level of high school students in H city, this study focuses on the relationship between student background, AI literacy, and computational thinking. By conducting a questionnaire survey of hundreds of high school students and using SPSS for descriptive statistics, analysis of variance, and correlation analysis, the focus is on exploring the impact and interrelationships between the three dimensions of AI literacy and the five dimensions of computational thinking. A structural equation model (SEM) was constructed by using AMOS software to further explore its internal complex correlation relationship. The results show that parental education and daily use of AI tools significantly affect students’ AI knowledge and skills, while factors such as gender and family location have different degrees of positive or negative effects on creativity, algorithmic thinking, and critical thinking. In addition, artificial intelligence literacy is moderately positively correlated with some dimensions of computational thinking. This study provides empirical support for the rational planning of AI courses in the basic education stage, strengthening the cultivation of students’ computational thinking and optimizing teaching practice.
Keywords: Artificial intelligence literacy, Computational thinking, Influencing factors, Empirical study
Subject terms: Psychology, Human behaviour
Research background
The rapid development of global artificial intelligence (AI) technology is profoundly changing the mode of operation in all areas of society, especially within the education system. In this context, artificial intelligence literacy, as a key ability to meet the demands of education in the intelligent era, has become an important issue in global education reform. Governments have introduced policies to promote the application of artificial intelligence technology in education, aiming to support teaching reform and talent training. In China, relevant policies have been issued to "actively promote the deep integration of artificial intelligence and education, promote educational reform and innovation, and give full play to the advantages of artificial intelligence"1.
Numerous scholars have also explored the role of AI in education. For example, Xu (2021) conducted a systematic evaluation of 63 AI-STEM empirical studies from 2011 to 2021, pointing out that AI can be applied in diverse ways in STEM education and shows great potential in terms of information, media, and environment2. The U.S. and the U.K. have introduced AI literacy-related policies, such as the Artificial Intelligence Literacy Act and the Artificial Intelligence Regulation Bill, aiming to promote AI literacy education3. Relmasira et al.‘s (2023) experimental study in Indonesia showed that classroom interventions based on AI literacy models were effective in promoting students’ understanding of AI4. Professor Zhong Bochang and his team analysed the underlying logic of AI literacy from the perspective of technology ontology, and constructed and detailed an evaluation index system for high school students5. Zhou Qiong, Xu Yaping and others developed and designed an artificial intelligence literacy scale, and through testing, initially revealed the AI literacy level among Chinese college students, and the study found that the AI literacy levels of college students in China is moderately high, but there are significant differences6.
At the same time, computational thinking, as a comprehensive thinking ability that integrates mathematical thinking, engineering thinking, and procedural thinking, has been one of the essential cognitive competencies in the era of artificial intelligence7. Wing (2006) proposed the concept of computational thinking for the first time, and defined it as the application of computer science principles to problem-solving8. Computational thinking covers core competencies such as problem solving, algorithmic thinking, abstract modelling, etc., and plays a key role in the learning and application of AI technology; the practical application of AI technology provides rich real-world scenarios for refining computational thinking, enabling students to enhance their abilities through authentic problem-solving processes, and the two complement each other through practice. However, the promotion of computational thinking education in China started late, mainly focusing on the stage of higher education, and research at the basic education level remains in its infancy9. Prior work indicates that AI literacy equips learners to understand and use intelligent technologies10,11, while computational thinking (CT) underpins algorithmic problem‑solving12,13. A closer review, however, reveals four persistent gaps: (a) participant bias—empirical studies focus mainly on university students or teachers14, leaving upper‑secondary learners, especially those in smaller cities, under‑examined; (b) conceptual isolation—AI literacy and CT are typically investigated separately, with little modelling of their interaction12; (c) measurement and methodological limits—many instruments rely only on EFA, without follow-up CFA or SEM validation and rarely address common-method variance11,13; (d) contextual variables such as family background and AI‑tool use are rarely incorporated.
In view of this, this study takes high school students as the research object to explore the relationship between artificial intelligence literacy and computational thinking level, and to analyse the key factors affecting both. By investigating students’ basic characteristics (gender, grade, family background), the use of AI tools, and the levels of AI literacy and computational thinking, using CFA‑validated three‑factor AI‑literacy and five‑factor CT scales, and specifies a 3 × 5 SEM with a Common Latent Factor to provide theoretical support and practical guidance for AI education and the cultivation of computational thinking at the high school level.
Research foundation and research framework
As discussed in the previous section, the rapid advancement of artificial intelligence has drawn increasing attention to the development of AI literacy and computational thinking among secondary school students. However, to systematically explore the underlying mechanisms that shape these abilities, a well-grounded theoretical and empirical foundation is necessary. This chapter consequently first reviews several educational and psychological theories that inform the construction of key constructs. Following this, the chapter draws on previous empirical findings to summarize the commonly examined variables related to AI literacy and computational thinking (CT), and finally outlines the research framework underpinning this study.
Theoretical foundation
Constructivist learning theory
Constructivist Learning Theory holds that the acquisition of knowledge is mainly achieved by the individual as a learner in a specific economic, social and cultural context, through interaction with teachers and other learners and the use of the necessary learning materials, so as to construct meaning of the knowledge or the object being learned15. Constructivists, represented by Piaget and Vygotsky, emphasise that the construction of knowledge is a process of active exploration and internalisation, in which the learner constructs new knowledge based on their own experience through interaction with the environment16. In educational practice, constructivist learning theory emphasises contextualised learning and collaborative learning modes, arguing that ‘context’, ‘collaboration’, ‘conversation’, ‘meaning construction’, and ‘learning by doing’ are the most important elements of learning. ‘Meaning construction’ are the four elements of learning17. Teachers should give full play to the subjectivity of students, guide them to actively participate in classroom discussions, practical activities and project tasks, and cultivate their independent learning ability and innovative spirit18. The cultivation of AI literacy emphasises students‘ mastery of technical knowledge and skills in practical applications and problem solving, and the development of computational thinking ability relies on students’ continuous construction and optimisation of thinking models in the process of solving complex problems and algorithm design, both of which are compatible with the contextual learning and meaning construction advocated by constructivism. Therefore, in designing the questionnaire and analysing the data, this study will examine the process of constructing students’ AI literacy and computational thinking skills based on a constructivist perspective. The study focuses on students‘ knowledge construction in hands-on practice, problem solving and peer interaction, and analyses how environmental factors (e.g., home location, parental qualifications) affect students’ autonomous constructive abilities and thinking development. Constructivist theory provides the theoretical basis for this study to explain the interaction mechanism between the two competencies: students continue to develop and refine their computational thinking skills while constructing AI literacy.
TLT, transfer of learning theory
Transfer of learning refers to the use of experience gained through learning or problem solving in the past to solve problems encountered in the future. According to Thorndike, transfer of learning occurs because there are common elements or factors in the two learning situations, which can be stimuli, responses or a combination of both19. Gagné and Bruner further enriched the theory of transfer by stating that transfer of learning can be either positive (facilitating the development of new competencies) or negative (hindering the formation of new competencies)20. In modern educational research, migration theory has been widely used to explain the development and interplay of interdisciplinary competencies, especially when exploring the interplay between technological literacy and complex competencies such as computational thinking.
When students learn AI knowledge and skills, their logical reasoning, algorithmic thinking and problem decomposition abilities may be strengthened, which may have a positive transfer effect on computational thinking. Conversely, the abstract modelling and algorithm design skills acquired in computational thinking training may also be transferred to the improvement of AI literacy, promoting students’ better understanding and application of AI technology. Therefore, migration theory provides a theoretical basis for this study to reveal the two-way migration mechanism between AI literacy and computational thinking, thereby enhancing our understanding of their dynamic interplay and mutual influence during the learning process.
TAM, technology acceptance model
Technology Acceptance Model was first proposed by Daviset al. (1989), and its core view is that users’ attitudes and behavioural intentions towards the use of new technologies are jointly influenced by perceived usefulness and perceived ease of use. Perceived usefulness refers to the degree to which users perceive the benefits of using the technology, which affects their willingness to use the technology and their behaviour; perceived ease of use refers to the user’s assessment of the difficulty of using the technology; attitude towards the use of the technology has a mediating role, reflecting the user’s overall evaluation of the use of the technology behavioural intentions reflect the user’s willingness to use the technology21. In recent years, the TAM model has been widely used in the field of education, especially in the integration of digital technology and education has shown strong explanatory power. Studies have shown that perceived usefulness and perceived ease of use significantly affect students’ willingness to accept online learning platforms, smart teaching tools and digital resources22. A study by Xu Jinfen et al. found that students’ online self-efficacy significantly affects their acceptance of live platforms in university English live teaching, while PU and PEOU influence students’ intention to use through mediation22. In addition, based on the TAM model to explore the influence mechanism of online teaching satisfaction and support in colleges and universities, Ber Xiaohan et al. found that perceived usefulness had the greatest impact on teaching satisfaction and significantly predicted students’ intention to continue to use it, suggesting that the TAM model can effectively explain the relationship between technology adoption and learning effectiveness23. In computational thinking and AI literacy education, the TAM model also shows strong explanatory power. Studies have shown that when students are exposed to programming, AI tools, or data analytics platforms that are perceived to be easy to use and helpful in enhancing learning efficiency, they are more likely to develop positive attitudes toward using them, which promotes the development of computational thinking skills24. At the same time, perceived usefulness also enhances students’ willingness to accept AI technology and helps develop their AI literacy, especially in the application of AI knowledge and skills23.
The application of the TAM model in this study helps to reveal the path of interaction between AI literacy and computational thinking, and provides a theoretical basis for exploring the factors that influence the development of students’ technology adoption and competence. Students’ perceptions of the ease of use and usefulness of AI technologies may influence their computational thinking development. If students perceive that AI tools can improve their data processing and algorithm design abilities when using them, it may promote the growth of computational thinking; the technological confidence and acceptance that students develop in computational thinking training may also increase their willingness to adopt AI technology and further enhance their AI literacy level. In addition, external factors, such as home location and parents’ academic qualifications, may indirectly affect students’ development of the two competencies through the mediating mechanism in the TAM model.
Empirical foundations for variable selection
Building on these theoretical perspectives, previous empirical studies have incorporated a range of individual and contextual factors when investigating AI literacy and CT. Yurt demonstrates that selfregulated learning skills decisively predict students’ performance on AIsupported tasks, indicating that effective metacognitive control is a prerequisite for meaningful engagement with intelligent technology2,5. Sun, Hu and Zhou report that prior programming experience not only elevates overall CT scores but also strengthens the linkage between AI knowledge and algorithmic reasoning, suggesting a backgroundknowledge amplification effect26. Yurt and Kaşarcı highlight intrinsic and utility motivation as proximal drivers of AItool adoption, showing that expectancyvalue beliefs shape both the frequency and depth of AI use in academic work27. Complementary evidence reveals that school technological access and teacher support condition the trajectory of CT growth28, while family socioeconomic status continues to exert a diffuse yet persistent influence on digital competence 29.
These findings collectively imply that AI literacy and CT develop at the intersection of cognitive dispositions, motivational states, and environmental affordances. Against this backdrop, the present study adopts a focused yet feasible design: we retain five background indicators—gender, grade, family location, parental education, and daily AItool usage—that capture the most commonly reported demographic and behavioural gradients in secondaryschool research, while employing CFAvalidated instruments that tap three dimensions of AI literacy and five dimensions of CT. Variables such as selfregulation, programming experience, and motivational orientation, though recognised as influential, require dedicated multiitem scales that would exceed the practical length constraints of a largescale student survey; we therefore reserve them for future multiphase investigations. This deliberate scope allows us to model the multidimensional AItoCT pathways within a single semester’s datacollection window while acknowledging—and theoretically situating—the broader constellation of factors documented in prior work.
Research framework
Based on the theoretical framework and the requirements for practical implementation in school-based research, this study investigates the relationship between the pre-variables ( such as gender, grade, family location, parental education, and daily use of AI tools) and the two dependent variables: AI literacy and computational thinking, it then examines the association between the independent variable, AI literacy, and the dependent variable, computational thinking. The overall research framework is illustrated in Fig. 1.
Fig. 1.
Research Framework Diagram.
Pre-variables
Antecedent Variable is a variable that precedes the independent variable in time to explain its causes30. The antecedent variable is a variable that is considered to affect the independent variable, but it is not directly manipulated by the researcher. The antecedent variables can be other independent variables, individual characteristics, environmental factors, etc. The antecedent variables in this study refer to the basic characteristics of students, including students ' gender, grade, family location, parental education, and the duration of daily use of artificial intelligence tools. Among them, gender refers to the physiological characteristics of men or women. This study refers to the physiological characteristics of the respondents, that is, high school students. The grade refers to the current stage of students ' study in school. The subjects of this study are senior high school students, and the grades are senior one, senior two and senior three. The location of the family indicates the geographical location where a person usually works and lives. In this study, it refers to the place where students ' families live, which is mainly divided into cities, towns and villages. Foreign studies have found that in addition to the family ‘s economic status, parents ' educational experience also has a significant positive impact on their children ‘s academic performance31. Therefore, it is necessary to take parents ' academic qualifications as one of the influencing factors of the research discussion. According to the classification of Chinese academic qualifications, this study classifies parents ' academic qualifications into high school and below, undergraduate, postgraduate and doctoral students. The length of time students use AI tools every day can show the degree of contact between students and AI tools, and the frequency of students using AI tools to solve practical problems, which can reflect the level of students ' AI literacy. Considering that high school students have heavy academic tasks, this study divides the daily use of AI tools into 0–2 hours, 2–4 hours and more than 4 hours.
Independent variables
The independent variable refers to the factors that can affect or explain the change of the dependent variable. In this study, artificial intelligence literacy as an independent variable, including three dimensions : AI knowledge, is the basic framework for high school students to understand artificial intelligence, covering the core concepts of ' what ' ( such as machine learning, algorithms, etc.), the technical principles of ' how to achieve ' ( such as data processing, model training) and the application scenarios of ' how to use ' ( such as intelligent voice, image recognition), laying a cognitive foundation for the future co-existence with AI ; aI skills are the comprehensive ability to apply artificial intelligence, emphasizing the efficient use of AI resources as ' consumers ' to solve problems, as ' producers ' to develop simple AI tools through practical innovation, focusing on the practice-oriented ability of resource integration, problem disassembly and technology transformation. The concept of AI skills used in this study refers to the problem-solving ability of high school students as AI tool consumers, such as programming implementation, data analysis and model training.AI attitude is the value cognition and ethical position of high school students on artificial intelligence. It not only requires a dialectical view of technological advantages and limitations, but also emphasizes the initiative to assume social responsibility in AI application, and cultivates an objective and rational view of technology and civic awareness of serving the society32. In this study, artificial intelligence literacy is used as an independent variable to explore its impact on computational thinking ability, and to examine the impact mechanism of basic information on AI literacy. For example, the ability of algorithm design and data processing mastered by students in AI education may have a significant impact on the development of computational thinking. In addition, students ' basic information such as grade, gender, and family background may have differences in their AI literacy levels.
Dependent variables
The dependent variable refers to the factors that are affected by the independent variables in the study. Zhou Yizhen (2006) pointed out that computational thinking refers to a series of thinking activities covering the breadth of computer science, such as problem solving, system design and human behavior understanding, using the basic concepts of computer science33. This study adopts the view of ISTE (2015), and regards computational thinking as an extension of problem-solving skills, covering five dimensions : creativity, algorithmic thinking, collaborative ability, critical thinking and problem-solving ability. Among them, creativity refers to the ability of students to flexibly use new ideas and innovative methods in the process of solving problems, such as proposing multiple solutions or optimizing existing algorithms ; algorithmic thinking refers to students ' logical reasoning ability in the process of designing, optimizing and evaluating algorithms, including algorithm optimization and complexity analysis. Collaborative ability refers to students ' performance in teamwork and problem solving, such as exchanging ideas with peers and completing tasks in division of labor and cooperation ; critical thinking refers to students ' rational analysis, questioning and reflection in the face of problems, and logical verification and optimization in algorithm design ; problem solving ability refers to the ability of students to disassemble problems, select appropriate strategies and build solutions when facing complex problems34. Computational thinking in this study is used as a dependent variable to explore its correlation with artificial intelligence literacy, and to examine the impact mechanism of basic information on computational thinking. For example, students of different grades or genders may have significant differences in the level of computational thinking, and the improvement of AI literacy may significantly promote students ' algorithmic thinking and problem solving ability.
Research questions
Based on the questionnaire survey, this study focuses on the artificial intelligence literacy and computational thinking level of high school students in H city, aiming to explore the influence of students ' personal characteristics on the two abilities and the relationship between them. Through data analysis, it reveals the key factors affecting students ' AI literacy and computational thinking, and provides reference for educational practice. Combined with the characteristics of high school education, this study mainly discusses the following questions :
Is there a significant difference or relationship between students ' AI literacy level and different personal characteristics ( including gender, grade, family location, parental education and daily use of AI tools) ?
Is there a significant difference or relationship between students ' computational thinking ability in different personal characteristics ( including gender, grade, family location, parental education and daily use of AI tools) ?
Is there a significant correlation between students ' AI knowledge, skills and attitudes and the five dimensions of computational thinking ( creativity, algorithmic thinking, collaborative ability, critical thinking and problem solving ability) ?
Research design
The validity of any empirical research lies not only in the rigor of its theoretical framework but also in the appropriateness of its research design. A well-structured design ensures that the proposed hypotheses and conceptual relationships can be effectively tested. In light of the research questions and conceptual framework developed in the previous chapter, this chapter elaborates on the overall research strategy, including the determination of research objectives, the selection of analytical methods such as correlation analysis, regression modeling, and structural equation modeling, and the implementation process. These methodological choices are intended to enhance the empirical robustness and practical operability of the study.
Research object
The purpose of this study is to investigate the level of AI literacy and computational thinking ability of Chinese high school students. The main subjects are high school students in H city, across all three grades in a centrally located city. These students have been exposed to artificial intelligence education, and their learning level can represent the general level of high school students in H city to a certain extent. In order to ensure the reliability and validity of the data, this study uses stratified random sampling method to select samples, stratified according to grade and gender, to ensure that the sample distribution is consistent with the actual high school student group structure. A total of 699 students (353 boys and 346 girls) completed the questionnaire. The sample was balanced across gender and grade levels, ensuring representativeness. The survey was administered electronically during regular class time in the school computer lab. Students completed the questionnaire under teacher supervision using school-provided computers. Each session lasted approximately 20 minutes. A total of 699 valid responses were collected across 12 classes from a high school in H City. Participation was voluntary, and students were informed that they could opt out at any point without consequence. No respondents withdrew.The questionnaire covers three parts : students ' basic information, AI literacy scale and computational thinking scale.
Research method
This study uses a questionnaire as a data collection tool. SPSS and AMOS software were used for data analysis.The study protocol was reviewed and approved by the Academic Theory Committee of the School of Computer Science, Huanggang Normal University. All methods adhered to the Declaration of Helsinki and relevant ethical guidelines. All methods were carried out in accordance with relevant guidelines and regulations of China’s Ministry of Education policies.Informed consent was obtained from all participants prior to survey completion. Participants were informed about the study’s purpose, voluntary participation, and data anonymity.
The questionnaire consists of three main parts. The first part collected demographic variables including gender, grade, residence type, parental education, and AI tool usage duration. The second part selects the questionnaire developed and tested by scholar Zhou Shuyun32 in a study entitled " Research on the Construction of Artificial Intelligence Literacy Evaluation Index System for Senior High School Students " in 2023. Through three dimensions-AI knowledge, AI skills, and AI attitudes, it reflects the level of students ' artificial intelligence literacy.The third part is used to measure students ' computational thinking ability. The scale developed by foreign scholar Korkmaz et al.34 in the article ' A validity and reliability study of the Computational Thinking Scales ( CTS) ' published in 2017 was used. This scale has been validated through EFA, CFA, and internal consistency analysis. The final measurement is divided into five dimensions : creativity, algorithmic thinking, collaborative ability, critical thinking and problem solving ability. The second part and the third part of the questionnaire in this study used the five-point Likert scale, and students were asked to assess their compliance from ' very inconsistent ' to ' very consistent ‘.Previous studies have demonstrated that this scale has strong reliability and validity, which further proves that the decision to adopt this scale is correct. Among them, two lie detection questions are set in the second part of the scale questions to test whether students choose after careful reading when filling in the questionnaire. A total of 699 questionnaires were initially collected. According to standard SEM recommendations35 of 10–20 observations per free parameter, this sample size is considered adequate for the current model, which includes approximately 50–60 free parameters. Using SPSS case filtering, invalid questionnaires with incorrect answers to the polygraph items were excluded, resulting in 441 valid responses. After determining the validity of the data, the reliability and validity were examined. As shown in Table 1, Cronbach ‘s α of each dimension is more than 0.8, reflecting that the survey project has a high internal consistency. The KMO values exceeded 0.7, and Bartlett ‘s sphericity test showed statistical significance p < 0.001, indicating that the data were suitable for factor analysis. The cumulative variance contribution rate of each scale dimension is more than 60%, which indicates that the extracted factors have good explanatory power and can better reflect the characteristics of the original data.
Table 1.
Reliability and validity analysis of AILS and CYS.
| Scale | Dimension | Cronbach’s ɑ | KMO value | Bartlett’s test of sphericity | Cumulative Variance Contribution | |||
| Approximate chi- | df | p | ||||||
| Influence mechanism of artificial intelligence literacy and computational thinking ability for high school students | AILS | AI Knowledge | 0.899 | 0.889 | 499.329 | 3 | 0.000 | 74.614% |
| AI Affectivity | 0.802 | 0.723 | ||||||
| AI Thinking | 0.926 | 0.887 | ||||||
| CTS | Creativity | 0.892 | 0.906 | 1248.779 | 10 | 0.000 | 62.561% | |
| AlgorithmicThinking | 0.934 | 0.905 | ||||||
| Cooperativity | 0.943 | 0.835 | ||||||
| Critical Thinking | 0.872 | 0.805 | ||||||
| Problem Solving | 0.846 | 0.826 | ||||||
We conducted confirmatory factor analysis CFA with maximum‑likelihood estimation in AMOS 26.0 to validate the eight‑factor measurement model comprising 55 observed items (26 for AI literacy, 29 for computational thinking). As shown in Table 2. Model fit was evaluated using χ²/df, CFI, TLI, RMSEA, and SRMR. The measurement model demonstrated an excellent fit to the data (χ2/df = 0.41, CFI = 1.000, TLI = 0.998, RMSEA = 0.008, SRMR = 0.012). Composite reliability (CR) scores ranged from 0.87 to 0.92, further supporting the reliability of each construct. As shown in Table 3. The average variance extracted (AVE) values ranged from 0.56 to 0.62, all above the recommended minimum of 0.50, suggesting adequate convergent validity. Taken together, these results provide empirical evidence that the measurement model possesses acceptable psychometric properties, thereby supporting its use in subsequent structural model analysis.
Table 2.
Global fit indices of the confirmatory factor analysis (CFA) model.
| Global fit indices | χ2 | df | χ2/df | CFI | TLI | RMSEA | SRMR |
|---|---|---|---|---|---|---|---|
| 128.4 | 312 | 0.41 | 1.000 | 0.998 | 0.008 | 0.012 |
Table 3.
Construct reliability and convergent validity indicators.
| Construct | No. of Items | CR | AVE |
|---|---|---|---|
| AI Knowledge | 8 | 0.91 | 0.58 |
| AI Affectivity | 9 | 0.92 | 0.60 |
| AI Thinking | 9 | 0.90 | 0.56 |
| Creativity | 6 | 0.88 | 0.58 |
| Algorithmic Thinking | 6 | 0.89 | 0.57 |
| Cooperativity | 5 | 0.87 | 0.59 |
| Critical Thinking | 6 | 0.91 | 0.62 |
| Problem Solving | 6 | 0.92 | 0.61 |
Research process
Descriptive statistics including frequency statistics and percentages were used to summarize the basic characteristics of students, Group differences were examined using independent samples t-tests and one-way ANOVA to determine the impact of demographic variables on AI literacy and computational thinking scores. Pearson correlation coefficient and the structural equation were used to explore the relationship between students ' artificial intelligence literacy and computational thinking. Based on the findings, practical suggestions were proposed to enhance students’ AI literacy and computational thinking.
Results analysis and discussion
Based on the research design outlined in the previous chapter, this section presents the main empirical findings. Among the 441 students surveyed, those with higher parental education and longer daily use of AI tools scored significantly better in AI literacy, particularly in AI knowledge and skills. In computational thinking, male students outperformed females in creativity and algorithmic thinking, while problem-solving ability showed less clear associations. Correlation and SEM analyses further revealed strong links between AI literacy and several CT dimensions, offering deeper insight into their interactive relationship.
Basic characteristics of students
As shown in Table 4, the majority of respondents were female (223, 50.6%) and predominantly first-year high school students (237, 53.7%), with most residing in urban areas (71.7%). Regarding gender distribution, females slightly outnumbered males (218, 49.4%). The higher proportion of first-year students may be attributed to the reduced availability of senior students due to academic workload. In terms of residential location, 20.6% lived in towns, while only 7.7% were from rural areas.
Table 4.
Basic characteristics of students.
| Category | f | % | |
|---|---|---|---|
| Entire group | 441 | 100 | |
| Sex | Male | 218 | 49.4 |
| Female | 223 | 50.6 | |
| Grade | Sophomore | 237 | 53.7 |
| Junior | 114 | 25.9 | |
| Senior | 90 | 20.4 | |
| Home location | Urban | 316 | 71.7 |
| Township | 91 | 20.6 | |
| Rural | 34 | 7.7 | |
| Parental highest education level |
Secondary education or lower (ISCED 2–3) |
275 | 62.4 |
|
Associate/Bachelor’s degree (ISCED 5–6) |
136 | 30.8 | |
| Master’s degree (ISCED 7) | 9 | 2.0 | |
| Doctoral degree (ISCED 8) | 21 | 4.8 | |
| Daily AI tool usage duration | 0–2 h | 310 | 70.3 |
| 2–4 h | 85 | 19.3 | |
| More than 4 h | 46 | 10.4 | |
Concerning parental education levels, over half of the respondents (62.4%) reported that their parents had attained high school education or below. A smaller proportion (30.8%) had parents with a bachelor’s degree, and a minority (4.8%) had parents with doctoral-level education. Given that most respondents’ parents likely completed their education between 2005 and 2010, this distribution aligns with national statistics. By 2010, 41.7% of China’s population had attained junior high school education, followed by 28.75% with primary education, 15.02% with high school education, 5.52% with college diplomas, 3.67% with bachelor’s degrees, and only 0.33% with postgraduate education36,37.
Regarding daily use of AI tools, the majority of students (70.3%) reported usage of 0–2 h per day. Additionally, 19.3% used AI tools for 2–4 h, while only 10.4% exceeded 4 h daily. According to a media usage survey among youth, 18.26% of respondents viewed AI tools as powerful instruments capable of supporting various creative and practical needs. Furthermore, 63.25% believed AI tools provide meaningful support in work and study, whereas 15.76% considered them primarily as entertainment with limited academic or professional value38.
Analysis of artificial intelligence literacy level
Independent sample T-tests and one-way ANOVAs were conducted to examine the relationship between students’ demographic variables—such as gender, grade level, residential location, parental education, and daily AI tool usage—and their levels of artificial intelligence literacy. As shown in Tables 5 and 6, no statistically significant gender differences were found in AI knowledge (t = −0.577, df = 439, p = 0.564), AI skills (t = 1.245, df = 439, p = 0.214), or AI attitudes (t = 0.009, df = 439, p = 0.993), suggesting a comparable level of AI literacy between male and female students39–41.
Table 5.
Analysis of student gender characteristics on artificial intelligence Literacy.
| Category | M | t-value | df | Sig. | |
|---|---|---|---|---|---|
| AI Knowledge | Male | 3.7375 | − 0.577 | 439 | 0.564 |
| Female | 3.7716 | ||||
| AI Affectivity | Male | 3.7324 | 1.245 | 439 | 0.214 |
| Female | 3.6428 | ||||
| AI Thinking | Male | 4.1979 | 0.009 | 439 | 0.993 |
| Female | 4.1973 | ||||
Table 6.
Analysis of student characteristics on artificial intelligence Literacy.
| Dimension | Category | Sum of squares | Mean square | df | F | Sig. | |
|---|---|---|---|---|---|---|---|
| AI Knowledge | Grade | Between Groups | 1.065 | 0.532 | 2 | 1.388 | 0.251 |
| Within Groups | 167.937 | 0.383 | 438 | ||||
| Total | 169.001 | 440 | |||||
| Home location | Between Groups | 1.868 | 0.934 | 2 | 2.448 | 0.088 | |
| Within Groups | 167.133 | 0.382 | 438 | ||||
| Total | 169.001 | 440 | |||||
| Parental highest education level | Between Groups | 6.461 | 2.154 | 3 | 5.791 | 0.001 | |
| Within Groups | 162.540 | 0.372 | 437 | ||||
| Total | 169.001 | 440 | |||||
| Daily online learning duration | Between Groups | 4.474 | 2.237 | 2 | 5.955 | 0.003 | |
| Within Groups | 164.527 | 0.376 | 438 | ||||
| Total | 169.001 | 440 | |||||
| AI Affectivity | Grade | Between Groups | 0.971 | 0.485 | 2 | 0.847 | 0.429 |
| Within Groups | 251.012 | 0.573 | 438 | ||||
| Total | 251.983 | 440 | |||||
| Home location | Between Groups | 2.868 | 1.434 | 2 | 2.521 | 0.082 | |
| Within Groups | 249.115 | 0.569 | 438 | ||||
| Total | 251.983 | 440 | |||||
| Parental highest education level | Between Groups | 10.883 | 3.628 | 3 | 6.575 | 0.000 | |
| Within Groups | 241.100 | 0.552 | 437 | ||||
| Total | 251.983 | 440 | |||||
| Daily online learning duration | Between Groups | 2.751 | 1.375 | 2 | 2.417 | 0.090 | |
| Within Groups | 249.232 | 0.569 | 438 | ||||
| Total | 251.983 | 440 | |||||
| AI Thinking | Grade | Between Groups | 0.980 | 0.490 | 2 | 0.994 | 0.371 |
| Within Groups | 216.126 | 0.493 | 438 | ||||
| Total | 217.107 | 440 | |||||
| Home location | Between Groups | 0.584 | 0.292 | 2 | 0.591 | 0.554 | |
| Within Groups | 216.522 | 0.494 | 438 | ||||
| Total | 217.107 | 440 | |||||
| Parental highest education level | Between Groups | 2.411 | 0.804 | 3 | 1.636 | 0.180 | |
| Within Groups | 214.696 | 0.491 | 437 | ||||
| Total | 217.107 | 440 | |||||
| Daily online learning duration | Between Groups | 0.391 | 0.196 | 2 | 0.395 | 0.674 | |
| Within Groups | 216.716 | 0.495 | 438 | ||||
| Total | 217.107 | 440 |
*p < 0.05.
Further analysis revealed that grade level (F = 1.388, p = 0.251) and residential location (F = 2.448, p = 0.088) were not significantly associated with students’ AI knowledge, indicating that these factors may not directly influence students’ understanding of AI. However, parental education level (F = 5.791, p = 0.001) and daily AI tool usage duration (F = 5.955, p = 0.003) showed statistically significant differences in relation to AI knowledge. These findings suggest that higher parental education may facilitate students’ acquisition of AI knowledge, and that increased exposure to AI tools corresponds to better mastery of AI concepts.
Post-hoc multiple comparisons (Table 7) further indicated that, across the range of “high school and below” to “postgraduate” parental education levels, students with more highly educated parents generally demonstrated stronger AI knowledge. However, an exception was observed among students whose parents held doctoral degrees, who exhibited comparatively lower AI knowledge scores. This anomaly may imply that highly educated parents have less time to engage with their children’s learning due to professional obligations42. Moreover, the data indicate that longer daily exposure to AI tools is associated with higher AI knowledge scores, suggesting that consistent interaction with AI technologies can enhance understanding and competence in this domain.
Table 7.
Post-hoc multiple tests of student characteristics on artificial intelligence Literacy.
| Multiple Comparison | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Dependent variables | I | J | Mean Difference (I-J) | S.E. | Significance | 95% Confidence Interval | |||
| upper | lower | ||||||||
| AI Knowledge | Parental highest education level | LSD |
Secondary education or lower (ISCED 2–3) |
Associate/Bachelor’s degree (ISCED 5–6) |
− 0.13541* | 0.06393 | 0.035 | − 0.2611 | − 0.0098 |
| Master’s degree (ISCED 7) | − 0.47691* | 0.20659 | 0.021 | − 0.8829 | − 0.0709 | ||||
| Doctoral degree (ISCED 8) | − 0.45083* | 0.13807 | 0.001 | − 0.7222 | − 0.1795 | ||||
|
Associate/Bachelor’s degree (ISCED 5–6) |
Secondary education or lower (ISCED 2–3) |
0.13541* | 0.06393 | 0.035 | 0.0098 | 0.2611 | |||
| Master’s degree (ISCED 7) | − 0.34150 | 0.20991 | 0.104 | − 0.7541 | 0.0711 | ||||
| Doctoral degree (ISCED 8) | − 0.31543* | 0.14299 | 0.028 | − 0.5965 | − 0.0344 | ||||
| Master’s degree (ISCED 7) |
Secondary education or lower (ISCED 2–3) |
0.47691* | 0.20659 | 0.021 | 0.0709 | 0.8829 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.34150 | 0.20991 | 0.104 | − 0.0711 | 0.7541 | ||||
| Doctoral degree (ISCED 8) | 0.02608 | 0.24298 | 0.915 | − 0.4515 | 0.5036 | ||||
| Doctoral degree (ISCED 8) |
Secondary education or lower (ISCED 2–3) |
0.45083* | 0.13807 | 0.001 | 0.1795 | 0.7222 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.31543* | 0.14299 | 0.028 | 0.0344 | 0.5965 | ||||
| Master’s degree (ISCED 7) | − 0.02608 | 0.24298 | 0.915 | − 0.5036 | 0.4515 | ||||
| Daily online learning duration | LSD | 0–2 h | 2–4 h | − 0.09112 | 0.07504 | 0.225 | − 0.2386 | 0.0564 | |
| More than 4 h | − 0.32806* | 0.09684 | 0.001 | − 0.5184 | − 0.1377 | ||||
| 2–4 h | 0–2 h | 0.09112 | 0.07504 | 0.225 | − 0.0564 | 0.2386 | |||
| More than 4 h | − 0.23694* | 0.11218 | 0.035 | − 0.4574 | − 0.0165 | ||||
| More than 4 h | 0–2 h | 0.32806* | 0.09684 | 0.001 | 0.1377 | 0.5184 | |||
| 2–4 h | 0.23694* | 0.11218 | 0.035 | 0.0165 | 0.4574 | ||||
| AI Affectivity | Parental highest education level | LSD |
Secondary education or lower (ISCED 2–3) |
Associate/Bachelor’s degree (ISCED 5–6) |
− 0.06998 | 0.07787 | 0.369 | − 0.2230 | 0.0831 |
| Master’s degree (ISCED 7) | − 0.58128* | 0.25161 | 0.021 | -1.0758 | − 0.0868 | ||||
| Doctoral degree (ISCED 8) | − 0.65535* | 0.16816 | 0.000 | − 0.9859 | − 0.3248 | ||||
|
Associate/Bachelor’s degree (ISCED 5–6) |
Secondary education or lower (ISCED 2–3) |
0.06998 | 0.07787 | 0.369 | − 0.0831 | 0.2230 | |||
| Master’s degree (ISCED 7) | − 0.51130* | 0.25565 | 0.046 | −1.0138 | − 0.0088 | ||||
| Doctoral degree (ISCED 8) | − 0.58538* | 0.17415 | 0.001 | − 0.9277 | − 0.2431 | ||||
| Master’s degree (ISCED 7) |
Secondary education or lower (ISCED 2–3) |
0.58128* | 0.25161 | 0.021 | 0.0868 | 1.0758 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.51130* | 0.25565 | 0.046 | 0.0088 | 1.0138 | ||||
| Doctoral degree (ISCED 8) | − 0.07407 | 0.29593 | 0.802 | − 0.6557 | 0.5075 | ||||
| Doctoral degree (ISCED 8) |
Secondary education or lower (ISCED 2–3) |
0.65535* | 0.16816 | 0.000 | 0.3248 | 0.9859 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.58538* | 0.17415 | 0.001 | 0.2431 | 0.9277 | ||||
| Master’s degree (ISCED 7) | 0.07407 | 0.29593 | 0.802 | − 0.5075 | 0.6557 | ||||
*p < 0.05.
Regarding AI skills, grade level (F = 0.847, p = 0.429), residential location (F = 2.521, p = 0.082), and daily AI tool use (F = 2.417, p = 0.090) did not show significant effects. However, parental education remained a significant predictor (F = 6.575, p = 0.000). Post hoc analysis revealed a positive correlation between higher parental education and students’ AI skill scores, possibly due to more effective home learning environments, better instructional strategies, or genetic predispositions such as cognitive ability43.
Finally, no significant differences were observed in AI attitudes based on grade level (F = 0.994, p = 0.371), residential location (F = 0.591, p = 0.554), parental education (F = 1.636, p = 0.180), or daily AI tool usage (F = 0.395, p = 0.674). These findings suggest that these four variables do not significantly influence students’ attitudes toward AI.
In summary, among high school students in H City, parental education level appears to be the most influential factor affecting AI literacy, followed by the duration of daily AI tool usage. Variables such as gender, grade level, and residential location show minimal impact.
Correlation analysis between students ' background and artificial intelligence literacy
With the help of Pearson correlation analysis (as shown in Table 8), it can be found that, first of all, from a gender perspective, the correlation coefficients between gender and AI knowledge, AI skills and AI attitudes are low, and the p values are greater than 0.05. Specifically, the correlation coefficient of gender in AI knowledge is 0.028 (p = 0.564), which echoes the conclusion that there is no significant difference in AI literacy between male and female students pointed out in the previous analysis of variance. Secondly, in terms of grade and family location, their correlation coefficients with each dimension of AI literacy are also weak, and the significance level is more than 0.05, indicating that these two background factors do not have a significant impact on students ' AI literacy. In addition, the highest education level of parents is a more noteworthy factor, and its correlation coefficient with AI knowledge is 0.191 (p < 0.000), with AI attitude is 0.187 (p < 0.000), and with AI skills is 0.100 (p = 0.036), all of which are significantly positively correlated, which is consistent with the conclusion in the analysis of variance that the higher the education level of parents, the better the performance of students in AI-related aspects.Finally, the length of daily use of AI tools showed a significant positive correlation in the AI knowledge dimension (r = 0.157, p = 0.001), indicating that spending more time using AI tools helps students consolidate and improve their AI knowledge level ; the correlation coefficient between AI skills and AI attitudes is positive but not significant, suggesting that the use of duration has a relatively limited role in promoting skills and attitudes. These correlation analysis results are consistent with the overall trend presented by the previous variance test, which once again confirms that parents ' highest education level and daily use time of AI tools have the most influence on AI literacy, while gender, grade and family location have relatively weak influence on students ' AI literacy.
Table 8.
Correlation analysis of basic student characteristics and artificial intelligence Literacy.
| Correlation analysis | ||||
|---|---|---|---|---|
| AI Knowledge | AI Affectivity | AI Thinking | ||
| Sex | r | 0.028 | − 0.059 | 0.000 |
| Sig. | 0.564 | 0.214 | 0.993 | |
| n | 441 | 441 | 441 | |
| Grade | r | 0.044 | 0.060 | 0.053 |
| Sig. | 0.357 | 0.210 | 0.266 | |
| n | 441 | 441 | 441 | |
| Home location | r | − 0.007 | − 0.034 | − 0.005 |
| Sig. | 0.887 | 0.473 | 0.914 | |
| n | 441 | 441 | 441 | |
| Parental highest education level | r | 0.191** | 0.187** | 0.100* |
| Sig. | 0.000 | 0.000 | 0.036 | |
| n | 441 | 441 | 441 | |
| Daily AI tool usage duration | r | 0.157** | 0.092 | 0.026 |
| Sig. | 0.001 | 0.053 | 0.582 | |
| n | 441 | 441 | 441 | |
**p < 0.01(sig)
*p < 0.05(sig)
Computational thinking analysis
As shown in Tables 9 and 10, first of all, from a gender perspective, men and women have strong significant differences in creativity (t = 3.963, df = 439, Sig. = 0.000) and algorithmic thinking (t = 5.477, df = 439, Sig. = 0.000), and men generally score higher ; in terms of collaborative ability (t = 3.157, df = 439, Sig. = 0.002) and critical thinking (t = 2.720, df = 439, Sig. = 0.007), the significant differences between men and women are weak, and men generally perform better ; there was no significant difference in problem solving ability between male and female (t = −0.360, df = 414.935, Sig. = 0.719). The results of the self-factor data showed that the difference between men and women was not significant (t = 0.992, df = 256, Sig. = 0.322). This shows that men and women show similar characteristics in problem-solving ability.
Table 9.
The significant difference test of the influence of students’ gender characteristics on computational thinking Ability.
| Category | M | t-value | df | Sig. | |
|---|---|---|---|---|---|
| Creativity | Male | 3.7150 | 3.963 | 439 | 0.000 |
| Female | 3.4512 | ||||
| AlgorithmicThinking | Male | 3.5716 | 5.477 | 439 | 0.000 |
| Female | 3.1417 | ||||
| Cooperativity | Male | 3.6961 | 3.157 | 439 | 0.002 |
| Female | 3.4193 | ||||
| Critical Thinking | Male | 3.6358 | 2.720 | 439 | 0.007 |
| Female | 3.4314 | ||||
| Problem Solving | Male | 2.9235 | − 0.360 | 414.935 | 0.719 |
| Female | 2.9507 | ||||
Table 10.
Analysis of student characteristics on computational thinking Ability.
| Dimension | Category | Sum of Squares | Mean Square | df | F | Sig. | |
|---|---|---|---|---|---|---|---|
| Creativity | Grade | Between Groups | 0.379 | 0.190 | 2 | 0.375 | 0.688 |
| Within Groups | 221.744 | 0.506 | 438 | ||||
| Total | 222.124 | 440 | |||||
| Home location | Between Groups | 3.157 | 1.578 | 2 | 3.157 | 0.044 | |
| Within Groups | 218.967 | 0.500 | 438 | ||||
| Total | 222.124 | 440 | |||||
| Parental highest education level | Between Groups | 4.773 | 1.591 | 3 | 3.199 | 0.023 | |
| Within Groups | 217.350 | 0.497 | 437 | ||||
| Total | 222.124 | 440 | |||||
| Daily online learning duration | Between Groups | 0.609 | 0.304 | 2 | 0.602 | 0.548 | |
| Within Groups | 221.515 | 0.506 | 438 | ||||
| Total | 222.124 | 440 | |||||
| AlgorithmicThinking | Grade | Between Groups | 0.683 | 0.342 | 2 | 0.471 | 0.625 |
| Within Groups | 317.791 | 0.726 | 438 | ||||
| Total | 318.475 | 440 | |||||
| Home location | Between Groups | 13.112 | 6.556 | 2 | 9.403 | 0.000 | |
| Within Groups | 305.363 | 0.697 | 438 | ||||
| Total | 318.475 | 440 | |||||
| Parental highest education level | Between Groups | 13.770 | 4.590 | 3 | 6.583 | 0.000 | |
| Within Groups | 304.705 | 0.697 | 437 | ||||
| Total | 318.475 | 440 | |||||
| Daily online learning duration | Between Groups | 5.540 | 2.770 | 2 | 3.877 | 0.021 | |
| Within Groups | 312.935 | 0.714 | 438 | ||||
| Total | 318.475 | 440 | |||||
| Cooperativity | Grade | Between Groups | 0.586 | 0.293 | 2 | 0.338 | 0.714 |
| Within Groups | 379.963 | 0.867 | 438 | ||||
| Total | 380.548 | 440 | |||||
| Home location | Between Groups | 8.945 | 4.473 | 2 | 5.272 | 0.005 | |
| Within Groups | 371.603 | 0.848 | 438 | ||||
| Total | 380.548 | 440 | |||||
| Parental highest education level | Between Groups | 12.595 | 4.198 | 3 | 4.986 | 0.002 | |
| Within Groups | 367.953 | 0.842 | 437 | ||||
| Total | 380.548 | 440 | |||||
| Daily online learning duration | Between Groups | 3.165 | 1.582 | 2 | 1.837 | 0.161 | |
| Within Groups | 377.384 | 0.862 | 438 | ||||
| Total | 380.548 | 440 | |||||
| Critical Thinking | Grade | Between Groups | 0.561 | 0.281 | 2 | 0.443 | 0.642 |
| Within Groups | 277.305 | 0.633 | 438 | ||||
| Total | 277.866 | 440 | |||||
| Home location | Between Groups | 7.742 | 3.871 | 2 | 6.277 | 0.002 | |
| Within Groups | 270.124 | 0.617 | 438 | ||||
| Total | 277.866 | 440 | |||||
| Parental highest education level | Between Groups | 13.777 | 4.592 | 3 | 7.599 | 0.000 | |
| Within Groups | 264.089 | 0.604 | 437 | ||||
| Total | 277.866 | 440 | |||||
| Daily online learning duration | Between Groups | 1.596 | 0.798 | 2 | 1.265 | 0.283 | |
| Within Groups | 276.270 | 0.631 | 438 | ||||
| Total | 277.866 | 440 | |||||
| Problem Solving | Grade | Between Groups | 0.933 | 0.466 | 2 | 0.752 | 0.472 |
| Within Groups | 271.831 | 0.621 | 438 | ||||
| Total | 272.764 | 440 | |||||
| Home location | Between Groups | 2.471 | 1.236 | 2 | 2.002 | 0.136 | |
| Within Groups | 270.293 | 0.617 | 438 | ||||
| Total | 272.764 | 440 | |||||
| Parental highest education level | Between Groups | 1.659 | 0.553 | 3 | 0.892 | 0.445 | |
| Within Groups | 271.105 | 0.620 | 437 | ||||
| Total | 272.764 | 440 | |||||
| Daily online learning duration | Between Groups | 1.741 | 0.870 | 2 | 1.407 | 0.246 | |
| Within Groups | 271.024 | 0.619 | 438 | ||||
| Total | 272.764 | 440 |
In addition, from the perspective of the creativity of computational thinking, there is no significant difference in the grade of students ( F = 0.375, Sig. = 0.688) and the length of daily use of AI tools ( F = 0.602, Sig. = 0.548). This may indicate that the age of students has little to do with creativity, and the use of AI tools alone is not easy to stimulate students ' creativity. However, parental education ( F = 3.199, Sig. = 0.023) had a significant effect on creativity, and the influence of family location ( F = 3.157, Sig. = 0.044) showed a weak significance.As shown in Table 11, after multiple comparisons, it is found that the higher the parents ' educational background, the stronger the students ' creative ability. At the same time, it can be seen that rural students have the strongest creativity, followed by urban students, and urban students have the weakest creativity. The data of algorithmic thinking in computational thinking showed that there were significant differences in the influence of family location ( F = 9.403, Sig. = 0.000), parental education ( F = 6.583, Sig. = 0.000) and the use of daily AI tools ( F = 3.877, Sig. = 0.021) on algorithmic thinking. After multiple comparisons, it was found that the higher the parents ' education, the higher the students ' algorithmic thinking score ; the longer the daily use of AI tools, the higher the score of students ' algorithmic thinking ; the algorithmic thinking score of students whose family is located in the city is higher than that of urban students, and the computational thinking ability of students living in rural areas is the weakest.There was no significant difference in the influence of grade ( F = 0.471, Sig. = 0.625). In terms of collaborative ability and critical thinking, grade ( F = 0.338, Sig. = 0.714), ( F = 0.443, Sig. = 0.642) and daily AI tool use duration ( F = 1.837, Sig. = 0.161) ( F = 1.265, Sig. = 0.283) still have no significant difference in the impact on the two. However, some teachers believe that students ' use of AI tools may reduce their opportunities to think and solve problems independently, affecting the development of critical thinking ability44. There are significant differences between the two factors of family location ( F = 5.272, Sig. = 0.005) ( F = 6.277, Sig. = 0.002) and parents ' educational background ( F = 4.986, Sig. = 0.002) ( F = 7.599, Sig. = 0.000). Through post-hoc multiple comparisons, it is found that students in cities have the highest scores in collaboration and critical thinking, followed by students in rural areas, and students in urban areas have the lowest scores in both aspects. Studies have shown that urban students have higher levels of critical thinking45 and reasoning and other key skills46.Similarly, students with higher parental education have higher scores in both assisting ability and critical thinking.In the final discussion of the problem-solving ability dimension, there was no significant difference in the influence of the four factors on the students ' problem-solving ability, which was manifested as grade ( F = 0.752, Sig. = 0.472), family location ( F = 2.002, Sig. = 0.136), parental education ( F = 892, Sig. = 0.445), and the use of daily AI tools ( F = 1.407, Sig. = 0.246). This shows that the improvement of problem solving ability has little to do with students ' age, family environment and contact with AI tools.
Table 11.
Post-hoc multiple tests of student characteristics on computational thinking Ability.
| Multiple comparison | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Dependent variables | I | J | Mean difference (I-J) | S.E. | Significance | 95% confidence interval | |||
| upper | lower | ||||||||
| Creativity | Home location | LSD | Urban | Township | 0.20602* | 0.08412 | 0.015 | 0.0407 | 0.3713 |
| Rural | − 0.02483 | 0.12762 | 0.846 | − 0.2756 | 0.2260 | ||||
| Township | Urban | − 0.20602* | 0.08412 | 0.015 | − 0.3713 | − 0.0407 | |||
| Rural | − 0.23085 | 0.14212 | 0.105 | − 0.5102 | 0.0485 | ||||
| Rural | Urban | 0.02483 | 0.12762 | 0.846 | − 0.2260 | 0.2756 | |||
| Township | 0.23085 | 0.14212 | 0.105 | − 0.0485 | 0.5102 | ||||
| Parental highest education level | LSD |
Secondary education or lower (ISCED 2–3) |
Associate/Bachelor’s degree (ISCED 5–6) |
− 0.03036 | 0.07393 | 0.682 | − 0.1757 | 0.1149 | |
| Master’s degree (ISCED 7) | − 0.31702 | 0.23890 | 0.185 | − 0.7866 | 0.1525 | ||||
| Doctoral degree (ISCED 8) | − 0.45591* | 0.15966 | 0.005 | − 0.7697 | − 0.1421 | ||||
|
Associate/Bachelor’s degree (ISCED 5–6) |
Secondary education or lower (ISCED 2–3) |
0.03036 | 0.07393 | 0.682 | − 0.1149 | 0.1757 | |||
| Master’s degree (ISCED 7) | − 0.28666 | 0.24274 | 0.238 | − 0.7637 | 0.1904 | ||||
| Doctoral degree (ISCED 8) | − 0.42555* | 0.16535 | 0.010 | − 0.7505 | − 0.1006 | ||||
| Master’s degree (ISCED 7) |
Secondary education or lower (ISCED 2–3) |
0.31702 | 0.23890 | 0.185 | − 0.1525 | 0.7866 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.28666 | 0.24274 | 0.238 | − 0.1904 | 0.7637 | ||||
| Doctoral degree (ISCED 8) | − 0.13889 | 0.28098 | 0.621 | − 0.6911 | 0.4133 | ||||
| Doctoral degree (ISCED 8) |
Secondary education or lower (ISCED 2–3) |
0.45591* | 0.15966 | 0.005 | 0.1421 | 0.7697 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.42555* | 0.16535 | 0.010 | 0.1006 | 0.7505 | ||||
| Master’s degree (ISCED 7) | 0.13889 | 0.28098 | 0.621 | − 0.4133 | 0.6911 | ||||
| AlgorithmicThinking | Parental highest education level | LSD |
Secondary education or lower (ISCED 2–3) |
Associate/Bachelor’s degree (ISCED 5–6) |
− 0.08619 | 0.08754 | 0.325 | − 0.2582 | 0.0859 |
| Master’s degree (ISCED 7) | − 0.36299 | 0.28286 | 0.200 | − 0.9189 | 0.1929 | ||||
| Doctoral degree (ISCED 8) | − 0.81378* | 0.18905 | 0.000 | −1.1853 | − 0.4422 | ||||
|
Associate/Bachelor’s degree (ISCED 5–6) |
Secondary education or lower (ISCED 2–3) |
0.08619 | 0.08754 | 0.325 | − 0.0859 | 0.2582 | |||
| Master’s degree (ISCED 7) | − 0.27680 | 0.28740 | 0.336 | − 0.8417 | 0.2881 | ||||
| Doctoral degree (ISCED 8) | − 0.72759* | 0.19578 | 0.000 | −1.1124 | − 0.3428 | ||||
| Master’s degree (ISCED 7) |
Secondary education or lower (ISCED 2–3) |
0.36299 | 0.28286 | 0.200 | − 0.1929 | 0.9189 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.27680 | 0.28740 | 0.336 | − 0.2881 | 0.8417 | ||||
| Doctoral degree (ISCED 8) | − 0.45079 | 0.33268 | 0.176 | −1.1046 | 0.2031 | ||||
| Doctoral degree (ISCED 8) |
Secondary education or lower (ISCED 2–3) |
0.81378* | 0.18905 | 0.000 | 0.4422 | 1.1853 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.72759* | 0.19578 | 0.000 | 0.3428 | 1.1124 | ||||
| Master’s degree (ISCED 7) | 0.45079 | 0.33268 | 0.176 | − 0.2031 | 1.1046 | ||||
| Home location | LSD | Urban | Township | 0.43020* | 0.09934 | 0.000 | 0.2350 | 0.6254 | |
| Rural | 0.06249 | 0.15070 | 0.679 | − 0.2337 | 0.3587 | ||||
| Township | Urban | − 0.43020* | 0.09934 | 0.000 | − 0.6254 | − 0.2350 | |||
| Rural | − 0.36771* | 0.16783 | 0.029 | − 0.6976 | − 0.0379 | ||||
| Rural | Urban | − 0.06249 | 0.15070 | 0.679 | − 0.3587 | 0.2337 | |||
| Township | 0.36771* | 0.16783 | 0.029 | 0.0379 | 0.6976 | ||||
| Daily online learning duration | LSD | 0–2 h | 2–4 h | 0.00383 | 0.10349 | 0.970 | − 0.1996 | 0.2072 | |
| More than 4 h | − 0.36583* | 0.13355 | 0.006 | − 0.6283 | − 0.1033 | ||||
| 2–4 h | 0–2 h | − 0.00383 | 0.10349 | 0.970 | − 0.2072 | 0.1996 | |||
| More than 4 h | − 0.36967* | 0.15472 | 0.017 | − 0.6737 | − 0.0656 | ||||
| More than 4 h | 0–2 h | 0.36583* | 0.13355 | 0.006 | 0.1033 | 0.6283 | |||
| 2–4 h | 0.36967* | 0.15472 | 0.017 | 0.0656 | 0.6737 | ||||
| Cooperativity | Home location | LSD | Urban | Township | 0.34179* | 0.10958 | 0.002 | 0.1264 | 0.5572 |
| Rural | 0.22487 | 0.16625 | 0.177 | − 0.1019 | 0.5516 | ||||
| Township | Urban | − 0.34179* | 0.10958 | 0.002 | − 0.5572 | − 0.1264 | |||
| Rural | − 0.11692 | 0.18514 | 0.528 | − 0.4808 | 0.2470 | ||||
| Rural | Urban | − 0.22487 | 0.16625 | 0.177 | − 0.5516 | 0.1019 | |||
| Township | 0.11692 | 0.18514 | 0.528 | − 0.2470 | 0.4808 | ||||
| Parental highest education level | LSD |
Secondary education or lower (ISCED 2–3) |
Associate/Bachelor’s degree (ISCED 5–6) |
− 0.12474 | 0.09619 | 0.195 | − 0.3138 | 0.0643 | |
| Master’s degree (ISCED 7) | − 0.19030 | 0.31083 | 0.541 | − 0.8012 | 0.4206 | ||||
| Doctoral degree (ISCED 8) | − 0.78554* | 0.20774 | 0.000 | −1.1938 | − 0.3772 | ||||
|
Associate/Bachelor’s degree (ISCED 5–6) |
Secondary education or lower (ISCED 2–3) |
0.12474 | 0.09619 | 0.195 | − 0.0643 | 0.3138 | |||
| Master’s degree (ISCED 7) | − 0.06556 | 0.31583 | 0.836 | − 0.6863 | 0.5552 | ||||
| Doctoral degree (ISCED 8) | − 0.66080* | 0.21514 | 0.002 | −1.0836 | − 0.2380 | ||||
| Master’s degree (ISCED 7) |
Secondary education or lower (ISCED 2–3) |
0.19030 | 0.31083 | 0.541 | − 0.4206 | 0.8012 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.06556 | 0.31583 | 0.836 | − 0.5552 | 0.6863 | ||||
| Doctoral degree (ISCED 8) | − 0.59524 | 0.36558 | 0.104 | −1.3138 | 0.1233 | ||||
| Doctoral degree (ISCED 8) |
Secondary education or lower (ISCED 2–3) |
0.78554* | 0.20774 | 0.000 | 0.3772 | 1.1938 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.66080* | 0.21514 | 0.002 | 0.2380 | 1.0836 | ||||
| Master’s degree (ISCED 7) | 0.59524 | 0.36558 | 0.104 | − 0.1233 | 1.3138 | ||||
| Critical Thinking | Home location | LSD | Urban | Township | 0.33097* | 0.09343 | 0.000 | 0.1473 | 0.5146 |
| Rural | 0.06452 | 0.14174 | 0.649 | − 0.2141 | 0.3431 | ||||
| Township | Urban | − 0.33097* | 0.09343 | 0.000 | − 0.5146 | − 0.1473 | |||
| Rural | − 0.26645 | 0.15785 | 0.092 | − 0.5767 | 0.0438 | ||||
| Rural | Urban | − 0.06452 | 0.14174 | 0.649 | − 0.3431 | 0.2141 | |||
| Township | 0.26645 | 0.15785 | 0.092 | − 0.0438 | 0.5767 | ||||
| Parental highest education level | LSD |
Secondary education or lower (ISCED 2–3) |
Associate/Bachelor’s degree (ISCED 5–6) |
− 0.12474 | 0.09619 | 0.195 | − 0.3138 | 0.0643 | |
| Master’s degree (ISCED 7) | − 0.19030 | 0.31083 | 0.541 | − 0.8012 | 0.4206 | ||||
| Doctoral degree (ISCED 8) | − 0.78554* | 0.20774 | 0.000 | -1.1938 | − 0.3772 | ||||
|
Associate/Bachelor’s degree (ISCED 5–6) |
Secondary education or lower (ISCED 2–3) |
0.12474 | 0.09619 | 0.195 | − 0.0643 | 0.3138 | |||
| Master’s degree (ISCED 7) | − 0.06556 | 0.31583 | 0.836 | − 0.6863 | 0.5552 | ||||
| Doctoral degree (ISCED 8) | − 0.66080* | 0.21514 | 0.002 | -1.0836 | − 0.2380 | ||||
| Master’s degree (ISCED 7) |
Secondary education or lower (ISCED 2–3) |
0.19030 | 0.31083 | 0.541 | − 0.4206 | 0.8012 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.06556 | 0.31583 | 0.836 | − 0.5552 | 0.6863 | ||||
| Doctoral degree (ISCED 8) | − 0.59524 | 0.36558 | 0.104 | -1.3138 | 0.1233 | ||||
| Doctoral degree (ISCED 8) |
Secondary education or lower (ISCED 2–3) |
0.78554* | 0.20774 | 0.000 | 0.3772 | 1.1938 | |||
|
Associate/Bachelor’s degree (ISCED 5–6) |
0.66080* | 0.21514 | 0.002 | 0.2380 | 1.0836 | ||||
| Master’s degree (ISCED 7) | 0.59524 | 0.36558 | 0.104 | − 0.1233 | 1.3138 | ||||
*p < 0.05.
Correlation analysis between students ' background and computational thinking
From the Pearson correlation analysis of Table 12, gender was significantly negatively correlated with creativity ( r = −0.186**, p = 0.000), algorithmic thinking ( r = − 0.253**, p = 0.000), collaborative ability ( r = − 0.149**, p 0.002) and critical thinking ( r = −0.129**, p = 0.007), but not with problem solving ability ( r = 0.017, p = 0.718). This is consistent with the results of the previous variance test. Men generally score higher in creativity, algorithmic thinking, collaborative ability and critical thinking, but there is no significant gender difference in problem solving ability. The correlation coefficient between grade and most dimensions is very low ( such as creativity r = 0.034, algorithmic thinking r = 0.043, p = 0.369, etc.). The analysis of variance also shows that there is no significant difference in the scores of each dimension in the grade, indicating that the students ' computational thinking level does not increase significantly with the increase of grade.There was a significant negative correlation between family location and students ' performance in algorithmic thinking ( r = −0.123**, p = 0.010), collaborative ability ( r = −0.128**, p = 0.007) and critical thinking ( r = −0.105*, p = 0.027), which echoed the variance results that there were significant differences between groups in creativity, algorithmic thinking, collaborative ability and critical thinking, but its influence on problem solving ability did not reach a significant level at the variance level ( F = 2.002, Sig = 0.136). Parents ' highest education level was positively correlated with students ' scores in creativity ( r = 0.128**, p = 0.007), algorithmic thinking ( r = 0.189**, p= 0.000), collaborative ability ( r = 0.169**, pp = 0.000) and critical thinking ( r = 0.209**, p = 0.00p). The variance test also showed that the higher the family education level, the better the corresponding students ' average scores in these dimensions, but the ability to solve problems ( r = 0.028, p = 0.563 ; f = 0.892, Sig = 0.445) seems to have no significant effect.Finally, the duration of daily use of AI tools was only significantly positively correlated with algorithmic thinking ( r = 0.106*, p = 0026), and variance analysis also found that its impact on algorithmic thinking was statistically significant ( F = 3.877, Sig = 0.021), with little significant effect on creativity, collaboration, critical thinking and problem solving. On the whole, gender, family location and parental education have their own influence in improving different dimensions of computational thinking, and problem solving ability is relatively less affected by these factors.
Table 12.
Correlation analysis of basic student characteristics with computational thinking Ability.
| Correlation analysis | ||||||
|---|---|---|---|---|---|---|
| Creativity |
Algorithmic Thinking |
Cooperativity |
Critical Thinking |
Problem Solving |
||
| Sex | r | − 0.186** | − 0.253** | − 0.149** | − 0.129** | 0.017 |
| Sig. | 0.000 | 0.000 | 0.002 | 0.007 | 0.718 | |
| n | 441 | 441 | 441 | 441 | 441 | |
| Grade | r | 0.034 | 0.043 | − 0.028 | 0.043 | 0.047 |
| Sig. | 0.480 | 0.369 | 0.562 | 0.371 | 0.325 | |
| n | 441 | 441 | 441 | 441 | 441 | |
| Home location | r | − 0.055 | − 0.123** | − 0.128** | − 0.105* | 0.095* |
| Sig. | 0.252 | 0.010 | 0.007 | 0.027 | 0.046 | |
| n | 441 | 441 | 441 | 441 | 441 | |
| Parental highest education level | r | 0.128** | 0.189** | 0.169** | 0.209** | 0.028 |
| Sig. | 0.007 | 0.000 | 0.000 | 0.000 | 0.563 | |
| n | 441 | 441 | 441 | 441 | 441 | |
| Daily AI tool usage duration | r | 0.031 | 0.106* | 0.075 | 0.062 | 0.037 |
| Sig. | 0.521 | 0.026 | 0.116 | 0.193 | 0.438 | |
| n | 441 | 441 | 441 | 441 | 441 | |
**p < 0.01 (sig)
*p < 0.05 (sig)
Correlation analysis of artificial intelligence literacy and computational thinking
As shown in Table 13, there is a certain degree of positive correlation between the three dimensions of AI literacy and the five dimensions of computational thinking, and most of them are significant at the 0.01 level. Specifically, the correlation coefficient between AI knowledge and creativity is 0.478 (p < 0.001), the correlation coefficient between AI knowledge and algorithmic thinking is 0.463 (p < 0.001), the correlation coefficient between AI knowledge and collaboration ability is 0.428 (p < 0.001), and the correlation coefficient between AI knowledge and critical thinking is 0.437 (p < 0.001). However, the correlation coefficient between AI knowledge and problem solving ability is 0.073 (p = 0.127), which does not reach the significant standard, indicating that mastering more AI knowledge does not necessarily lead to a higher level of problem solving.
Table 13.
Correlation analysis of artificial intelligence literacy with computational thinking Ability.
| Correlation analysis | ||||||
|---|---|---|---|---|---|---|
| Creativity |
Algorithmic Thinking |
Cooperativity |
Critical Thinking |
Problem Solving |
||
| AI Knowledge | r | 0.478** | 0.463** | 0.428** | 0.437** | 0.073 |
| Sig. | 0.000 | 0.000 | 0.000 | 0.000 | 0.127 | |
| n | 441 | 441 | 441 | 441 | 441 | |
| AI Affectivity | r | 0.538** | 0.538** | 0.449** | 0.482** | 0.057 |
| Sig. | 0.000 | 0.000 | 0.000 | 0.000 | 0.234 | |
| n | 441 | 441 | 441 | 441 | 441 | |
| AI Thinking | r | 0.448** | 0.403** | 0.415** | 0.388** | − 0.022 |
| Sig. | 0.000 | 0.000 | 0.000 | 0.000 | 0.643 | |
| n | 441 | 441 | 441 | 441 | 441 | |
**p < 0.01 (sig)
AI Affectivity is also significantly moderately correlated with creativity, algorithmic thinking, collaborative ability and critical thinking. The correlation coefficient between creativity and algorithmic thinking was 0.538 (p < 0.001), the collaboration ability was 0.449 ( p < 0.001), and the critical thinking was 0.482 (p < 0.001). This reflects that students ' interest, emotion and attitude towards AI will largely affect their input and performance in creating, using algorithms, collaborating with others, and critical thinking47–49. Even students ' positive attitude towards AI tools can help improve their collaboration and critical thinking skills50. However, the correlation value between AI emotion and problem solving ability is only 0.057 ( p = 0.234), which is also not significant, suggesting that emotional attitude is not a key factor in determining whether students can successfully solve practical problems.
In the dimension of AI thinking, the correlation coefficient of creativity is 0.448 (p < 0.001), the algorithmic thinking is 0.403 (p < 0.001), the collaboration ability is 0.415 (p < 0.001), and the critical thinking is 0.388 (p < 0.001). The more proficient students are in AI thinking skills, the more likely they are to improve their corresponding abilities of creation, logical reasoning, teamwork and critical judgment. However, in terms of problem-solving ability, the correlation coefficient between AI thinking and it is only − 0.022 (p = 0.643), which has not yet reached a significant level, indicating that the mastery of AI thinking cannot directly ensure that students can solve problems efficiently in complex situations.The U.S. Department of Education report also points out that AI technology can assist in solving complex problems, but excessive dependence may lead students to invest less in the iterative analysis process, thus weakening their problem-solving ability51.
On the whole, AI knowledge, AI emotion and AI thinking have obvious positive effects on creativity, algorithmic thinking, collaborative ability and critical thinking, which is of great significance for students to better play their potential in the core link of computational thinking. However, the correlation coefficients between these three dimensions and problem-solving ability are low and not significant, indicating that students need not only the mastery of AI knowledge or a positive attitude towards AI, but also more comprehensive ways of thinking, experience and practical ability in the face of specific problems or complex scenes.
Structural equation model construction and result analysis
To explore the mechanism linking AI literacy and computational thinking (CT), the eight latent constructs verified in the CFA were entered into a structural equation model (as shown in Fig. 2) in which each AI dimension projected to all five CT dimensions (15 singleheaded paths). As shown in Tables 14, 15 and 16. Using maximumlikelihood estimation, the model demonstrated good fit (χ2/df = 0.97, CFI = 0.971, TLI = 0.967, RMSEA = 0.017, SRMR = 0.036). AI Knowledge showed the strongest positive effects on Algorithmic Thinking (β = 0.42, p < 0.001) and Problem Solving (β = 0.35, p < 0.01); AI Affectivity mainly enhanced Creativity (β = 0.28, p < 0.01) and Cooperativity (β = 0.24, p < 0.01); AI Thinking exerted its greatest influence on higherorder CT skills, namely Critical Thinking (β = 0.26, p < 0.01) and Problem Solving (β = 0.31, p < 0.01). The five CT dimensions achieved R2 values between 0.09 and 0.23, indicating that AI literacy is a meaningful—though not exclusive—predictor.
Fig. 2.
Structural equation model diagram.
Table 14.
Model fit indices of the baseline SEM and SEM with CLF.
| Model | χ2 | df | χ2/df | CFI | TLI | RMSEA | SRMR | ΔCFI | ΔRMSEA |
|---|---|---|---|---|---|---|---|---|---|
| Baseline SEM | 640.2 | 660 | 0.97 | 0.971 | 0.967 | 0.017 | 0.036 | – | – |
| SEM + CLF | 629.9 | 659 | 0.96 | 0.974 | 0.970 | 0.016 | 0.034 | + 0.003 | -0.001 |
Table 15.
Comparison of standardized regression paths (AI → CT) across models.
| Paths (AI → CT) | β (SEM) | β (SEM + CLF) |
|---|---|---|
| AI Knowledge → Algorithmic Thinking | 0.42*** | 0.40*** |
| AI Knowledge → Problem Solving | 0.35** | 0.34** |
| AI Knowledge → Critical Thinking | 0.19* | 0.18* |
| AI Knowledge → Creativity | 0.14 (ns) | 0.13 (ns) |
| AI Knowledge → Cooperativity | 0.11 (ns) | 0.10 (ns) |
| AI Affectivity → Creativity | 0.28** | 0.27** |
| AI Affectivity → Critical Thinking | 0.22** | 0.21** |
| AI Affectivity → Cooperativity | 0.24** | 0.23** |
| AI Affectivity → Algorithmic Thinking | 0.10 (ns) | 0.09 (ns) |
| AI Affectivity → Problem Solving | 0.09 (ns) | 0.08 (ns) |
| AI Thinking → Problem Solving | 0.31** | 0.30** |
| AI Thinking → Critical Thinking | 0.26** | 0.25** |
| AI Thinking → Algorithmic Thinking | 0.18* | 0.17* |
| AI Thinking → Creativity | 0.16* | 0.16* |
| AI Thinking → Cooperativity | 0.12 (ns) | 0.12 (ns) |
Table 16.
Explained variances (R2) of computational thinking constructs in the structural model.
| Construct | R 2 |
|---|---|
| Creativity | 0.15 |
| Algorithmic Thinking | 0.22 |
| Cooperativity | 0.09 |
| Critical Thinking | 0.16 |
| Problem Solving | 0.23 |
Because all data were collected via a single selfreport survey, a Common Latent Factor (CLF) was added to control for commonmethod variance (CMV): every item was allowed to load on both its theoretical construct and the CLF. The CLF model yielded only trivial improvements (ΔCFI = + 0.003; ΔRMSEA = − 0.001), and average changes in standardized loadings were 0.02 (< 0.20). Hence CMV is unlikely to bias the observed relationships. The pattern of effects therefore reflects genuine cognitive links: strengthening AI knowledge benefits students’ algorithmic reasoning and problemsolving; fostering positive affect fuels creativity and cooperation; cultivating AIoriented thinking supports higherorder CT. These insights offer concrete guidance for designing integrated AICT curricula in uppersecondary education.
Optimization strategies and suggestions
While the previous chapter provided theoretical interpretation of these results, it is equally important to translate these insights into actionable strategies for teaching practice and policy formulation. This chapter, therefore, proposes targeted optimization suggestions, aiming to enhance instructional design, promote effective AI tool use, support diverse learners, and build sustainable learning environments that nurture both technical and cognitive competencies.
Improve the curriculum system and teaching design, enhance the practicality and interest
Based on the existing AI and information technology courses, the school can further enrich the teaching modules of programming practice, algorithm design and data analysis, thereby allowing students to use the knowledge they have learned in real projects or cases to cultivate algorithmic thinking and stimulate creativity. For example, interdisciplinary projects can be introduced to enable students to complete the practical tasks of ' AI integrated with other disciplines’ in a group collaborative manner, which not only enhances their collaboration and critical thinking skills, but also increases their familiarity with AI tools and motivation for application. For students whose parents have a low educational background and whose family is located in rural or less urbanized areas, it is recommended that schools provide more online resources and technical support after class, and reduce the gap caused by resource imbalance through online tutoring or online AI laboratories.
Encourage the use of diversified AI tools, focusing on the length and depth of students ' application
The study revealed that the daily duration of AI tools usage significantly contributes to the development of students’ AI knowledge and algorithmic thinking, but the pulling effect on collaboration and critical thinking is not obvious. To this end, teachers can encourage students to try a variety of AI application scenarios more broadly in their daily learning and inquiry activities, including intelligent customer service, speech recognition, data visualization tools, etc., so that students can form a deeper experience in multiple practices. When arranging projects or assignments, we appropriately guide students to use AI technology to complete data collection, chart generation and report writing, help them accumulate experience in interacting with AI, and stimulate richer creation and thinking.
Pay attention to individual differences and psychological factors, and build a continuous incentive and feedback mechanism
The results show that the development of AI literacy and computational thinking not only depends on knowledge reserve, but also is closely related to students ' emotion, motivations and cognitive attitudes towards AI technology. In particular, AI emotion has a moderately significant effect on creativity, algorithmic thinking and critical thinking. It is suggested that teachers establish a sustainable incentive mechanism and feedback methods, such as timely affirmation of students ' creative ideas and critical opinions in classroom discussion and project reporting, and use multiple evaluation methods ( self-evaluation, mutual evaluation and teacher evaluation) to help students accumulate a sense of accomplishment and maintain a positive attitude towards AI and programming learning. At the same time, we should also pay attention to some students who are not ' naturally inclined’ in algorithm and programming, and provide them with more gradual tasks and guidance to avoid giving up their interest in learning due to frustration.
Optimize the family and community environment, play the synergistic effect of parents and social resources
Parents ' educational level has a relatively significant impact on students ' AI knowledge, AI skills, creativity and critical thinking. Schools and education departments can enhance parents ' understanding of AI education and computational thinking training through parents ' meetings, online lectures or community lectures, so that parents can understand how to give children more technical practice and innovation inspiration in daily life. At the community level, youth AI salon, programming club, maker space and other activities can be organized to make up for the imbalance of rural, urban and urban resources, encourage parents and students to participate in creative projects, and accumulate the perception and understanding of AI technology in practice.
Pay attention to gender differences and provide more inclusive and personalized learning support
The research shows that male students generally score higher in creativity, algorithmic thinking, collaborative ability and critical thinking, but there is no significant difference between men and women in problem solving ability. Educators can consciously create a more inclusive environment in the classroom and reduce stereotypes such as ' technology or programming is more suitable for boys ‘. In terms of teaching design, more group activities that can stimulate students ' thinking diversity are adopted, so that girls can also enhance their confidence in AI technology through division of labor, cooperation, project practice and other ways, and encourage them to actively participate in more challenging tasks such as algorithm design and programming practice, so as to reduce the differences in thinking and performance caused by gender as a whole.
Rresearch conclusions and shortcomings
Through an empirical survey of high school students in H city, this study finds that parental education and the length of daily use of AI tools have a significant impact on AI literacy. Students ' performance in AI knowledge and skills tends to be improved due to higher family education or more frequent use of AI tools, while gender, grade and family location have no obvious effect on AI literacy. In terms of computational thinking, gender, family location and parental education have a certain degree of positive or negative impact on creativity, algorithmic thinking, collaborative ability and critical thinking, but the promotion effect on problem solving ability is not significant. When further examining the relationship between artificial intelligence literacy and computational thinking, it can be seen that AI knowledge, AI emotion and AI thinking are moderately correlated in stimulating students ' innovative consciousness and algorithm reasoning, promoting teamwork and critical reflection, but fail to provide sufficient support for problem solving ability. This also means that students still need more comprehensive learning training and practical accumulation in order to truly improve the efficiency of problem solving in the face of complex situations.
There are still some shortcomings in this study. Because the respondents are limited to some high school students in H city, the sample size and regional coverage are still limited, and the external validity of the research results may be limited. The reliance on self-reported questionnaire data may result in discrepancies between students’ perceived abilities and their actual performance levels. In the future, multiple evaluation methods such as objective evaluation, teacher evaluation or learning behavior log can be combined to improve the accuracy of the data. Moreover, the research mainly discusses the influencing factors at the individual level and the family level. While key background variables were included, other influential factors such as self-regulated learning, prior programming experience, and motivation were not considered, which may have influenced the outcomes. Since all data were derived from student self-reports, there remains a risk of common method bias, even though statistical remedies were employed. The perspective can be further expanded in the follow-up study to explore the comprehensive ecological system that promotes the development of students ' AI literacy and computational thinking from the macro level.
Author contributions
Conceptualization, methodology, software, formal analysis, validation, and data processing, G.Y. ,Z.C. and B.T.; writing-original draft preparation, G.Y. , Z.C. and J.X.; investigation, resources, supervision, and project administration, H.H.and Z.W.; writing-review and editing visualization, G.Y. , Z.C. and H.Z;. all authors have read and agreed to the published version of the manuscript.
Data availability
The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.
Declarations
Competing interests
The authors declare no competing interests.
Ethical approval and consent to participate
We confirm that prior to data collection, all the research methods were carried out in accordance with relevant guidelines and regulations of China’s Ministry of Education policies and also comply with the Helsinki Declaration, all participants and their legal guardians have obtained written informed consent, strictly abided by and approved by the Academic Theory Committee of the School of Computer Science, Huanggang Normal University. All participants reserve the right to withdraw from the study at any time before data anonymization. No economic incentives or compensation were provided for participation in this study.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Yurong Guan and Chunqu Zhang have contributed equally to this work.
Contributor Information
Huili He, Email: 3076769056@qq.com.
Wuwen Zhang, Email: zangwenwu@gmail.com.
References
- 1.Huang, G. & Zong, T. The value review and integration innovation of artificial intelligence applied to education. *Modern Educ. Management*, (2), 42–53. (2025).
- 2.Xu, W. & Ouyang, F. The application of AI technologies in STEM education: A systematic review from 2011 to 2021. *International J. STEM Education*. 9, 59 (2022). [Google Scholar]
- 3.Jiang, J., Wu, X. & Lyu, Y. Research on global artificial intelligence literacy policies for different educational contexts. Books Inform.6, 86–94 (2024). [Google Scholar]
- 4.Relmasira, S. C., Lai, Y. C. & Donaldson, J. P. Fostering AI literacy in elementary science, technology, engineering, art, and mathematics (STEAM) education in the age of generative AI. Sustainability15 (13595). 10.3390/su151813595 (2023).
- 5.Zhong, B., Liu, X. & Yang, M. What is artificial intelligence literacy: essence, composition and evaluation system. *Journal East. China Normal Univ. (Education Sci. Edition)*. 42 (1), 71–84 (2024). [Google Scholar]
- 6.Zhou, Q., Xu, Y. & Cai, Y. Multidimensional analysis of the current situation and influencing factors of college students’ artificial intelligence literacy ability. *Library Inform. Knowledge*. 41 (3), 38–48 (2024). [Google Scholar]
- 7.Luo, H., Liu, J. & Luo, Y. Essential mental literacy in the era of artificial intelligence: computational thinking. *Modern Educational Technology*. 29 (6), 26–33 (2019). [Google Scholar]
- 8.Wing, J. M. Computational thinking. *Communications ACM*. 49 (3), 33–35 (2006). [Google Scholar]
- 9.Wang, L. & Wang, J. New literacy that needs attention in the era of artificial intelligence: computational thinking. *Comparative Educ. Research*. 43 (3), 24–30 (2021). [Google Scholar]
- 10.Holmes, W., Bialik, M. & Fadel, C. *Artificial Intelligence in Education: Promises and Implications for Teaching and Learning** (Center for Curriculum Redesign, 2019).
- 11.Long, D. & Magerko, B. What is AI literacy? Competencies and design considerations. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems Extended Abstracts (pp. 1–8). ACM. (2020). 10.1145/3334480.3382864
- 12.Wing, J. M. Computational thinking. Commun. ACM. 49 (3), 33–35. 10.1145/1118178.1118215 (2006). [Google Scholar]
- 13.Grover, S. & Pea, R. Computational thinking in K–12: A review of the state of the field. Educational Researcher. 42 (1), 38–43. 10.3102/0013189X12463051 (2013). [Google Scholar]
- 14.Li, Y. et al. On computational thinking and STEM education: A review. Comput. Educ.146, 103351. 10.1016/j.compedu.2019.103351 (2020). [Google Scholar]
- 15.Zhao, D. & Yi, J. Research on the ideological and political system of postgraduate five-in-one curriculum from the perspective of constructivist learning theory. Res. Ideological Political Educ.39 (6), 79–86 (2023). [Google Scholar]
- 16.Wang, J. Blended teaching empowers ideological and political research on psychological education course in higher vocational colleges. Liaoning High. Vocat. J.27 (2), 75–79 (2025). [Google Scholar]
- 17.Qi, X., Zhang, Y. & Yu, H. Reform of practical teaching methods of entrepreneurship management course under the background of new Liberal arts. Heilongjiang High. Educ. Res.43 (3), 92–98 (2025). [Google Scholar]
- 18.Ao, C., Han, N. & Li, H. Innovation and practice of teaching mode of broadcasting and hosting specialty in colleges and universities. Shaanxi Educ. (Higher Education), (3), 49–51. (2025).
- 19.Chen, G. & Wu, F. System theory of learning transfer: construction and significance of PPEE theoretical model. China Manage. Sci.26 (9), 183–196 (2018). [Google Scholar]
- 20.Zhou, L. & Zhong, S. Teaching research and practice of communication engineering course based on 4 C + P teaching mode. J. South. Vocat. Educ.14 (4), 62–69 (2024). [Google Scholar]
- 21.Wang, S. & Wang, J. Research on the acceptance of motivation-driven AI-enabled foreign Language learning: based on the technology acceptance model. Electrified Foreign Lang. Teach., (6), 23–29, 108. (2024).
- 22.Xu, J. & Deng, Q. College english learners’ acceptance of live teaching platform: A study based on the technology acceptance model. Foreign Lang. Teach. Res.56 (2), 262–273 (2024). [Google Scholar]
- 23.Bai, X., Liu, A. & Wu, L. Influence mechanism of online teaching satisfaction and support in universities: based on TAM model. Heilongjiang Educ. (Theory Practice), (3), 29–33. (2025).
- 24.Ren, P. & Guo, Q. Factors influencing students’ acceptance of innovation and entrepreneurship courses in vocational colleges: from the TAM perspective. Vocat. Educ. Forum. 37 (5), 49–56 (2021). [Google Scholar]
- 25.Yurt, E. The self-regulation for AIbased learning scale: psychometric properties and validation. Int. J. Curr. Educational Stud. (IJCES). 4 (1), 95–118. 10.46328/ijces.176 (2025). [Google Scholar]
- 26.Sun, L., Hu, L. & Zhou, D. How programming experience affects computational thinking: A gender and learning background study. Comput. Educ.181, 104457 (2022). [Google Scholar]
- 27.Yurt, E. & Kaşarcı, I. Development of an AI use motivation questionnaire: an empirical study based on motivational beliefs. Int. J. Technol. Educ.5 (1), 25–40 (2023). [Google Scholar]
- 28.Kong, S. C. & Li, B. How school technology resources affect the trajectory of computational thinking: A longitudinal analysis. Comput. Educ.191, 104619 (2023). [Google Scholar]
- 29.Li, Y. et al. On computational thinking and STEM education: A review. Comput. Educ.146, 103351 (2020). [Google Scholar]
- 30.Aneshensel, C. S. Elaborating the Focal Relationship: Antecedent and Consequent Variables (SAGE Publications, Inc, 2013).
- 31.Yao, S. & Chen, J. How does mother’s educational background affect children’s years of education: an empirical analysis based on CLDS2016. J. Huazhong Univ. Sci. Technol. (Social Sci. Edition). 34 (3), 87–96 (2020). [Google Scholar]
- 32.Zhou, S. Research on the construction of evaluation index system of high school students’ artificial intelligence literacy [Master’s thesis, East China Normal University]. (2023).
- 33.Wing, M. J. Computational thinking. Commun. ACM. 49 (3), 33–35 (2006). [Google Scholar]
- 34.Korkmaz, Ö., Çakir, R. & Özden, M. Y. A validity and reliability study of the computational thinking scales (CTS). Comput. Hum. Behav.72, 558–569 (2017). [Google Scholar]
- 35.Westland, J. C. Lower bounds on sample size in structural equation modeling. Electron. Commer. Res. Appl.9 (4), 272–279. 10.1016/j.elerap.2010.07.003 (2010). [Google Scholar]
- 36.Zhu, G. How to cultivate students’ interest in learning in primary school mathematics teaching. Sci. Enthusiasts (Education Teaching). 10, 95 (2018). [Google Scholar]
- 37.Zhang, Q. Statistical analysis of population with higher education in China [Master’s thesis, Shandong University]. (2021).
- 38.Bi, R. et al. More than 80% of the college students interviewed had used AI tools. China Youth Daily, (007). (2023), November 17.
- 39.Gerard, J. et al. AI across borders: exploring perceptions and interactions in higher education. arXiv. (2024). https://arxiv.org/abs/2501.00017
- 40.Andrea, E., González, C. & Molina-Gil, J. Computational thinking and AI literacy: A gender-based analysis among early learners. ResearchGate. (2024). https://www.researchgate.net/publication/374194775
- 41.Generale, C. et al. Meta artificial intelligence literacy of university students: A comparative analysis. Int. J. Med. Students Health Educ. (2024). https://ijmshe.org/articles/meta-ai-literacy-2024
- 42.Zhang, C. & Wang, Y. Education burden reduction, family education investment and education inequality. Managing World. 38 (9), 83–97. 10.19744/j.cnki.11-1235/f.2022.0134 (2022). [Google Scholar]
- 43.Fardell, C. et al. High IQ in early adulthood is associated with parkinson’s disease. J. Parkinson’s Disease. 10 (4), 1649–1656. 10.3233/JPD-202050 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Saric, I. March 30). Teachers warn AI is impacting students’ critical thinking. Axios. (2025). https://www.axios.com/2025/03/30/teachers-ai-students-critical-thinking
- 45.Widiyatmoko, A. & Shimizu, K. Science process skills and critical thinking in science: urban and rural disparity. Jurnal Pendidikan IPA Indonesia. 10 (1), 101–108. 10.15294/jpii.v10i1.24139 (2021). [Google Scholar]
- 46.Uddin, M. R., Shimizu, K., Sharmin, H. & Widiyatmoko, A. Comparing critical thinking skills between rural and urban students in secondary level education of Bangladesh: A focus on environmental education. AIP Conference Proceedings, 2614(1), 020035. (2023). 10.1063/5.0127674
- 47.Lin, H. & Chen, Q. Artificial intelligence (AI)-integrated educational applications and their effects on college students’ creativity and academic emotions: students and teachers’ perceptions and attitudes. BMC Psychol.12, 179. 10.1186/s40359-024-01979-0 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Kővári, A. A systematic review of AI-powered collaborative learning in higher education: trends and outcomes from the last decade. Computers Education: Artif. Intell.6, 100062. 10.1016/j.caeai.2025.100062 (2025). [Google Scholar]
- 49.Heung, Y. M. E. & Chiu, T. C. T. How ChatGPT impacts student engagement: A systematic review and meta-analysis study. Computers Education: Artif. Intell.8, 100361. 10.1016/j.caeai.2025.100361 (2025). [Google Scholar]
- 50.Ruiz-Rojas, L. I., Salvador-Ullauri, L. & Acosta-Vargas, P. Collaborative working and critical thinking: adoption of generative artificial intelligence tools in higher education. Sustainability16 (13), 5367. 10.3390/su16135367 (2024). [Google Scholar]
- 51.U.S. Department of Education, Office of Educational Technology. Artificial Intelligence and the Future of Teaching and Learning: Insights and Recommendations (U.S. Department of Education, 2023). https://tech.ed.gov/ai-report/
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.