Artificial Intelligence and Data Literacy in Rural Schools’ Teaching Practices: Knowledge, Use, and Challenges

Highlights

What are the main findings?

• More than half of the surveyed teachers reported having a moderate to high level of knowledge of AI. However, the practical application of AI in the classroom is limited.
• The main concerns of teachers center on ethical implications, academic integrity (plagiarism and cheating), and a possible reduction in students’ critical thinking skills.

What are the implications of the main findings?

• AI should be implemented in ways that optimize resource allocation, reduce the digital divide, promote teacher mobility between urban and rural areas, refine subject structures through university partnerships, and improve training mechanisms for rural educators.
• It is essential to develop skills in both the use of AI tools and data management, ultimately contributing to improving learning outcomes and promoting educational equity within rural communities. Data literacy is critical to taking advantage of AI applications in education.

Abstract

This study explores the implementation of artificial intelligence (AI) and data literacy in rural Catalan schools by analyzing teacher knowledge, use, and perceptions. Data were collected from a representative sample of teachers at these schools to examine their understanding of AI and data literacy, how they utilize these technologies, and their perspectives on their applications. The results indicate that although over half of the teachers reported moderate to high AI knowledge, classroom implementation remains limited. Teachers primarily employed AI for text generation and content detection, with less frequent use of video generation or simulations. Common applications included lesson planning and material creation. Concerns centered on ethical implications, academic integrity, and a potential reduction in students’ critical thinking skills. This study reveals a moderate level of AI and data literacy knowledge among teachers in Catalan rural schools, contrasting with its limited practical application in the classroom. Teachers mainly use AI for text generation and content detection. Regarding data literacy, teachers demonstrated knowledge but lacked practical skills. These findings reveal a disconnect between theoretical AI knowledge and its practical application in the classroom, emphasizing the need for enhanced training and support to facilitate effective AI integration in education

1. Introduction

Current research in artificial intelligence (AI) within education reveals a deepening understanding of its capacity to personalize learning, offer targeted assistance, and improve teaching methods. A key area of investigation involves using AI to forecast student achievement and detect students who may be struggling. Researchers are exploring how AI algorithms can analyze student data to generate insights that facilitate early interventions and customized learning trajectories. This strategy seeks to enhance learning outcomes by adapting educational experiences to individual needs and proactively addressing potential obstacles. Crucially, scholars underscore the necessity of fairness and bias analysis in these AI models to guarantee equitable educational opportunities for all students, thereby mitigating potential biases that could reinforce existing disparities.

Several core principles support the effective implementation of AI in education. The importance of secure computing in AI education is emphasized, ensuring data privacy and responsible technology use (Bissadu & Hossain, 2024). The value of personalized feedback is also highlighted, using AI to customize support for individual students’ needs (Booth et al., 2024). Additionally, collaborative instructional design is advocated, exploring how AI can enable collaboration between educators and learners in developing engaging learning experiences (Parsons & Curry, 2024).

The increasing adoption of various AI tools and technologies demonstrates a growing dedication to harnessing these advancements in education. These include machine-learning tools, like Weka Chichekian and Benteux (2022), for introducing AI concepts to students, high-dosage tutoring systems that deliver personalized feedback through AI-driven discourse analytics, and the investigation of AI’s potential in automating and improving instructional design processes.

Moreover, studies highlight various educational practices that effectively integrate AI (C.-H. Chen & Chang, 2024; Heeg & Avraamidou, 2023; González-Calatayud et al., 2021). Examples delve into the specific applications of AI in science education, including the types of applications, the teaching content, and the impacts on science learning and teaching. They include AI-supported game-based learning, which uses AI to boost engagement and motivation; problem-based learning, enhanced through co-design with teachers, where AI facilitates collaborative problem solving; and dynamic transitions between individual and collaborative learning, facilitated by AI orchestration tools, allowing for adaptive learning experiences tailored to both individual and group needs.

In the specific context of rural schools, in Catalonia, they are defined as small, local, and public institutions, deeply rooted in their territories and acting as a key driver of local life. These schools face particular challenges, such as the difficulty in adapting to established regulations, limitations in resources and infrastructure, and the need for specific training for teachers (Manzano et al., 2018; i Franco, 2019). Although educational technology is increasingly prevalent in rural public schools, there is a significant gap in our understanding of how teachers utilize these tools. Specifically, we lack data on the frequency of use, teachers’ perceptions of effectiveness, and the obstacles they encounter when integrating technology into their instruction (Kormos & Wisdom, 2021).

The digital divide is a reality, particularly for rural communities, where access to quality education can be limited (Kormos & Wisdom, 2021). That means the integration of artificial intelligence (AI) within rural education presents both unique challenges and promising opportunities. This integration occurs within a complex context characterized by specific structural and pedagogical considerations. These include adapting curricula to reflect and reconstruct rural identities (as opposed to simply mirroring urban-centric models), navigating the complexities of multigrade teaching and fluctuating student populations, and fostering school autonomy within broader rural school zones (i Franco, 2019). Successfully leveraging AI in this environment necessitates a nuanced understanding of these contextual factors alongside teachers’ perceptions and attitudes toward the technology.

A critical element of this framework is the exploration of teacher attitudes toward AI. Although studies like Santos et al. (2024) and L. Chen et al. (2020) suggest that socio-geographical factors, like school location, gender, and area of specialization, do not significantly influence interest in AI, they highlight the crucial roles of perceived social impact and anxiety about AI’s behavior and potential dominance in shaping negative attitudes. Importantly, this research also indicates that teachers are capable of interacting and communicating with AI, suggesting that targeted training programs can address anxieties and foster positive engagement. This underscores the need to move beyond simplistic assumptions about teacher resistance and delve into the specific concerns and knowledge gaps that influence their perceptions of AI. Addressing these misconceptions and fostering positive interactions with AI are crucial for successful integration.

Furthermore, the existing literature highlights broader challenges within rural education that AI integration must address. Some researchers (Hongli & Leong, 2024; Yao, 2020) point to critical issues, such as unequal resource distributions, teacher shortages, subject-matter structural imbalances, and deficits in professional expertise. These challenges are amplified in rural settings and necessitate a strategic approach to AI implementation. They suggest that AI can be leveraged to optimize resource allocation, bridge the digital divide, promote teacher mobility between urban and rural areas, refine subject-matter structures through university partnerships, and enhance training mechanisms for rural educators. By addressing these systemic challenges, AI can contribute to greater educational equity and improve the quality of education for rural students. Therefore, the approach of this project aligns with this framework and posits that effective AI integration in rural education requires not only addressing teacher attitudes and anxieties but also strategically deploying AI to mitigate existing structural inequalities and empower rural educators. This combined approach can maximize the potential of AI to enhance learning outcomes and promote educational equity in rural communities.

Thus, the convergence of generative AI and LA in the Catalan education system presents a scenario of great potential but also of complex challenges, specifically in rural centers. These technologies can be crucial tools for meaningful learning and personalization, creating new possibilities for rapid diagnosis and intervention, the creation of adapted content, knowledge construction, and personalized curricular adaptations. By integrating personalized learning platforms, intelligent tutoring systems, and automated management tools, AI can provide tailored learning paths, reduce teachers’ administrative workloads, and improve teaching efficiency. Zhang and Leong (2024) emphasize the importance of lightweight AI models, edge computing, and differential privacy technology to overcome infrastructural limitations and data privacy concerns in these environments.

However, the lack of preparation, knowledge, and specific guidelines on their application and effective evaluation in various active methodologies has resulted in fragmented and, in some cases, ineffective or ethically questionable implementation. Furthermore, the lack of research in this area hinders the development of best practices and policies, particularly in the Catalan rural context, where educational needs are diverse and specific. However, although the research is focused on Catalonia, the results are expected to be transferable and adaptable to other areas of Europe. Solving this problem is imperative not only for the quality and effectiveness of education but also for preparing the next generation to live and thrive in a digital and data-driven world. In this sense, this study investigates the use of artificial intelligence (AI) and data literacy in teaching practice, exploring correlations between sociodemographic variables and AI usage, perception, and knowledge and examining the relationship between AI use and data literacy among teachers.

2. Materials and Methods

This study investigates the use of artificial intelligence (AI) and data literacy in teaching practice, exploring correlations between sociodemographic variables and AI usage, perception, and knowledge and examining the relationship between AI use and data literacy among teachers. The study employed a quantitative approach, utilizing survey data collected from secondary school teachers.

The objectives of this analysis are to

O1: Identify areas of artificial intelligence (AI) that are underutilized in teaching practice through descriptive and frequency analyses.

Research Question: Which specific applications of AI are the least frequently used by teachers in their teaching practices?

O2: Explore the correlations between the use, perception, and knowledge of AI in educational practice variables.

Research Questions: Is there a correlation between teachers’ perceived benefits of AI and their actual use of AI tools in the classroom? Does a higher level of AI knowledge among teachers correlate with a greater frequency of AI tool usage? Is there a relationship between teachers’ perceptions of AI’s ease of use and their reported knowledge of AI?

O3: Explore the relationship between AI use and data literacy among teachers.

Research Question: Does a higher level of data literacy among teachers correlate with greater adoption and use of AI tools in their teaching practices?

The study sample consisted of 141 teachers. A representative sample of secondary school teachers from schools in municipalities with a population of less than 2000 inhabitants. All the participating teachers provided informed consent prior to their inclusion in the study, ensuring ethical research practices and protecting the participants’ confidentiality. To gather participants, we employed a recruitment strategy involving collaboration with high schools, securing prior institutional consent. Subsequently, a snowball sampling methodology was implemented to connect with teachers from each participating school.

The statistical validation of the teacher questionnaire involved data preprocessing, including removing incomplete data and applying label encoding. Reliability analysis using Cronbach’s alpha (0.946) and polychoric alpha (0.942) indicated excellent internal consistency. Split-half reliability analysis showed high correlation (0.95) and corrected reliability (0.98), confirming excellent reliability. Construct validity was assessed using Bartlett’s test and KMO (0.891), determining three factors: IA use, IA in teaching and knowledge, and concern. Correlation analysis revealed coherent relationships among IA knowledge, usage frequency, and concern. The construct analysis was deemed as valid because of thematic coherence among grouped questions and consistent correlations.

To minimize the data loss and impacts on its structure, the following methods were used to handle null values. Prior to addressing the null values, label encoding was performed. In this process, each variable category was assigned a numerical value from 0 to i-1.

Data from items Q1 to Q10 were encoded in alphabetical order. The mode was used to replace null values, meaning that empty fields were substituted with the most frequently occurring category. For the remaining items, encoding followed a logical ascending order, and the variable mean was used to replace null values.

The survey instrument included items assessing knowledge, use, perception, and data literacy related to AI in education. Specific variables and their corresponding survey items are detailed below:

Knowledge measured knowledge of various AI applications, including general AI knowledge, classroom application, AI tools, detecting AI-generated content, and generating content (text, code, images, audio, video, music), prompt engineering, and simulation/prediction with AI.

Use assessed current uses of AI tools and resources in teaching practice; the frequency of AI tool use, including sub-items for content detection and generation (conversation, text, images, audio, video, music, presentation, educational content, programming code, simulations/predictions); and the purpose of AI tool use, covering activity planning, material creation, feedback, management, and data analysis. Experience in using AI for teaching and learning was also assessed.

Perception measured concern regarding AI, including concerns about fact verification, privacy, ethics, and skill assessment, and the perceived suitability of AI in various educational areas, such as STEM, social sciences, languages, and performing arts.

Data Literacy assessed skill levels in data handling.

Variable	Scale	Item
Knowledge	None Low Moderate High Expert	AI in general Applying AI in the classroom AI tools Detecting content generated with AI Generating textual content with AI Generating programming code with AI Generating images with AI Generating audio with AI Generating videos with AI Generating music with AI Prompt engineering Simulation and prediction with AI
Perception	Not worried at all A little worried Moderately worried Very worried Extremely worried	Reduced fact checking Reduced critical thinking Inaccuracies and errors Bias and discrimination Digital divide Academic ethics and integrity, honesty, and misconduct (plagiarism and cheating) Privacy issues Assessment of skills and knowledge Legality
	Strongly disagree Disagree Neutral or unsure Agree Strongly agree	STEM (science, technology, engineering, and mathematics) Social sciences (history, economics, geography, etc.) Language arts (reading, writing, and literature) Foreign languages (English, French, German, etc.) Physical education Performing arts (music, theater, and dance) Not applicable to any area
Use	Never Almost never Sometimes Often Constantly	AI detection Conversation generation Text generation Image generation Audio generation Video generation Music generation Presentation generation Educational content generation Programming code generation Simulations and predictions
		Plan/prepare lessons, activities, or learning experiences Create materials and resources for students Teach lessons to students Plan assignments and assessments Provide feedback/assess students Communicate with families Communicate and collaborate with colleagues Communicate with students Assist in management tasks (emails, reports, etc.) Research tasks to improve teaching practice Generate/contrast ideas or information Generate reports Analyze educational data
	No experience Limited experience Some experience Experienced Highly experienced	Do you have experience in the use of artificial intelligence applied to teaching and learning?
Data literacy	I do not feel capable. I feel capable of accomplishing this task with help. I feel capable of accomplishing this task in a basic way. I feel capable of accomplishing this task in its entirety. I carry out this task in my professional practice. I carry out this task in my professional practice and am capable of helping and teaching others.	I use data from digital learning environments (e.g., learning activities in the virtual classroom, applications, WhatsApp for teaching, or other tools) to identify the learning problems of students and groups to whom I teach. I use data from digital learning environments to solve problems in educational practice, with the student and the groups to whom I teach. I record the solutions to educational problems identified through data through reports, memoirs, diaries, or other tools. I use different data sources to extract quality assessments of my own teaching practice. I use assessment, content transmission, or visualization tools that allow me to work with unprocessed data to extract automatic and elaborated results (e.g., the use of WhatsApp for teaching, Google/Microsoft Forms, Moodle, Quiz, and Google Data Studio). I understand how to analyze, manage, and add data to my activity as a teacher. I make modifications to the design of learning activities according to problems detected in students through data generated in digital learning environments (for example, performance statistics in activities, logins, activity, or inactivity on the platform, and the level of participation in forums). I make modifications to my teaching practice in the development of learning activities based on the types of problems detected through the data (for example, very long times in the development of activities or on the contrary, impulsiveness, and scores below what was expected). I take into account data on student performance and learning processes in digital learning environments when implementing learning activities (e.g., prioritizing certain activities over others to reinforce content in groups or individually). I take into account data obtained through student activity in digital learning environments when evaluating teaching activities (e.g., detecting needs for improvement in activities through data). I use tables, graphs, and visualizations to represent and communicate data generated in digital learning environments. I test hypotheses about students’ learning processes or my own teaching practice through the representation of data generated in digital learning environments (e.g., detect changes in students’ performance and/or interest through the visualization of tables or graphs for the comparison of performance averages, times, and increases or decreases in participation in forums). I evaluate patterns and trends in the learning process through elements of visualization or the representation of data generated in digital learning environments (e.g., graphical representation of trends in the decrease in student performance averages when long statements are introduced in exercises). I synthesize and explain different sets of data on learning through elements of visualization or the representation of data (tables, graphs, diagrams, etc.) generated in digital learning environments. I take into account the ethical aspects of the processes of visualization, representation, and dissemination of data generated in digital learning environments. I know the privacy policies and legal implications of the handling and disclosure of data in the educational context of the classroom. I know the organizational chart, roles, and internal processes of the educational institution relating to the handling and disclosure of data in the educational context of the classroom (e.g., school data protection officer and school data protection guide). I monitor students’ performance through data generated in digital learning environments. I generate added value from students’ performances and needs through the handling of data generated in digital learning environments (e.g., through students’ grades, I generate information on the causes of learning difficulties). I make adjustments to the representation of contents, in my own teaching practice and in the monitoring of students based on the results obtained in the handling of data generated in digital learning environments. I understand the individual, social, and cultural learning contexts in which data for decision making have been extracted. I promote students to carry out self-learning, -supervision, and -monitoring practices based on data generated in digital learning environments (e.g., virtual classrooms and applications that enable activities).

Following data exploration and preprocessing in Python, descriptive statistical analysis was performed to understand the sample’s variable distributions. Given the observed non-normal distribution of all the variables (as confirmed by Shapiro–Wilk and Kolmogorov–Smirnov tests, p < 0.001), a non-parametric approach to correlation analysis was deemed as appropriate. Specifically, Spearman’s rank correlation was employed to assess relationships between variables, as it does not rely on assumptions of normality.

3. Results

3.1. Descriptive Statistical Analysis

The analysis of the teachers’ demographic data reveals featured trends regarding the adoption of AI in education. In terms of gender, a slight majority of women (60.8%) is observed. Regarding age, most teachers who use or are interested in using AI are between 36 and 60 years old. Finally, the analysis of teaching specialties shows that the areas of mathematics, science, and technology (35.8%) and languages (29.8%) lead in the incorporation of AI, pointing to a greater affinity with this technology in disciplines that are based on data analysis, problem solving, and communication.

Additionally, the educational levels where teaching stands out among teachers are the fourth year of ESO (42.3%), followed by first and third years of ESO (20.9% and 20.3%, respectively). A total of 29.9% of the sample is a part of the management team, and a minority of 3.6% is a part of the technological coordination team. A total of (89.7%) of the participants work full time, and the remaining figure is part time. A total of 47.5% of the participants have more than 16 years of experience. A total of 28.8% of the participants have between six and fifteen years of experience, although 3.5% of the participants have less than one year of experience.

Knowledge about AI

Regarding knowledge about AI, it is observed that 53.7% of the teachers report a moderate or a high (9.8%) level of knowledge of AI, representing half of the sample. In contrast, its application in the classroom is mostly low (31.7%) or non-existent (4.9%).

A substantial portion of teachers (19.5%) reported moderate knowledge of AI tools, with an additional 8.3% claiming high or expert proficiency. Specifically, they felt the most knowledgeable in generating textual content with AI (43.9% moderate knowledge and 17% high or expert level) and generating images with AI (17.1% moderate and 13.2% high or expert level).

Knowledge of specific applications, such as AI-generated content detection, code generation, AI audio generation, AI video generation, AI music generation, prompt engineering, and AI simulations and predictions, has low usage rates, indicating that mastery of advanced AI skills is less common (Figure 1).

Uses of AI

Most teachers who use AI in their practice (frequently or daily) do so for specific tasks, such as conversation generation, text generation, and content detection, at 14.6%, 17.1%, and 12.3%, respectively. Although there is significant adoption of AI in creation and verification tasks, other applications, such as video generation and simulations, have less use (2.4%), reflecting untapped potential in AI tools for teaching.

Lesson planning (22% often or constantly) and creating materials (17.1% often or constantly) for students are the most common purposes for using AI, which represent a significant plurality. This pattern suggests that teachers are primarily integrating AI to optimize the design and development of pedagogical content. Teachers prioritize the pedagogical applications of AI over its potential for research and communication with stakeholders (families, colleagues, and students), as evidenced by lower ratings in these latter areas (Figure 2).

Perceptions of AI

The most prominent concerns are focused on ethical aspects and academic integrity (plagiarism and cheating), which are indicated by 24.4% (extremely concerned) and 58.5% (very concerned) of the respondents, and in the reduction in critical thinking (17.1% extremely concerned and 61% very concerned). These items reflect that, especially regarding its impacts on students’ skills and academic honesty.

Teachers associate AI with technical and scientific disciplines. Most teachers see AI as a suitable tool for STEM areas (80.5%), while its applicability in subjects such as physical education is low (7.3% agree with its suitability).

Data literacy

The responses for the data literacy questions indicate that half of the sample (between 48% and 66%) feel capable of performing tasks with assistance or in a basic way. However, the proportion decreases when it comes to more advanced skills related to the operation of the center or a part of the data-handling process, such as

• Knowing the organizational chart, roles, and internal processes of the educational institution related to the handling and dissemination of data in the educational context of the classroom (e.g., the data protection officer of the center and data protection guide of the center). More than half are not very capable, and 14.6% do not feel capable.
• Monitoring students’ performance through data generated in digital learning environments. More than half feel not very capable, and 14.7% do not feel capable.
• Generating added value from students’ performances and needs through the handling of data generated in digital learning environments (e.g., through student grades, generating information about the causes of learning difficulties). More than half feel not very capable, and 14.7% do not feel capable.

This indicates that although there is a base of knowledge in data handling, the lack of practical skills limits the ability of teachers to apply data literacy effectively in the classroom. The variability in data literacy (high coefficients of variation) also suggests a dependence on external factors, such as access to training in data technologies.

3.2. Correlation Analysis

Given the non-normal distribution of the sample, Spearman’s rank correlation was selected. This type of correlation, like Kendall’s tau, is appropriate for non-normal, ordinal, or Likert-scale data, as is the case herein. Both tests measure the strength and direction of the relationship between variables. Spearman is useful for detecting monotonic relationships.

Frequency and Purpose of AI Use

The analysis of the frequency of AI use and the associated educational purpose reveals significant correlations between the specific uses of AI tools and pedagogical objectives, as can be observed in the correlation matrices.

The main correlations are regarding the generation of programming code, showing high and positive correlations with the purposes of lesson planning and preparation (r = 0.730) and the creation of materials and resources for students (r = 0.785). These results suggest that teachers who frequently use AI for code generation also tend to employ it in the preparation of educational content and teaching resources.

Similarly, the generation of images using AI correlates positively and significantly with both lesson planning and preparation (r = 0.683) and the creation of materials and resources for students (r = 0.712). This indicates that generating visual content with AI is a practice commonly linked to these pedagogical objectives.

Overall, most items in AI use and its purpose show positive and significant correlations between 0.20 and 0.60, suggesting moderate relationships. Although various AI usage practices are positively associated with different pedagogical objectives, these relationships are especially strong in the cases of code generation and image creation; in the processes of planning/preparing lessons, activities, or learning experiences; and in creating materials and resources for students.

Relationship between AI Knowledge and Frequency of Use in Teaching

A significant correlation was found between AI knowledge and frequency of use in the teaching context, with Pearson coefficients ranging between 0.2 and 0.6 for most items (p < 0.001). This relationship suggests that greater AI knowledge is associated with a higher frequency of use in various applications.

Some noteworthy correlations include

• Image generation: The correlation between the frequency of AI use for image generation and knowledge about generating textual content with AI was moderately high (r = 0.626).
• Audio generation: A strong correlation (r = 0.689) was observed between knowledge about generating images with AI and the frequency of AI use for audio generation.

Relationship between the Level of AI Use and Perception of AI

Negative and significant correlations were identified between the level of AI use, which ranges from not knowing any tools to using them daily, and the perception of concern regarding AI in different contexts, such as a reduction in fact-checking, legality issues, and skill assessment. These negative correlations suggest that greater AI use is associated with a lower level of concern about its effects on the educational field. The negative correlations between the level of AI use and the perception of concerns range between −0.1 and −0.2, indicating significant, albeit weak, relationships (p < 0.05). The negative correlation between the frequency of AI use and concern about the assessment of skills and knowledge stands out, with a coefficient of −0.2 (p < 0.01). This suggests that teachers who use AI more frequently tend to show less concern regarding its implications for skill assessment.

Data Literacy and Frequency of AI Use

Positive and significant correlations were found between the level of AI use, which ranges from not knowing tools to using them daily, and the level of competence in data handling in educational settings. In general, correlations between AI use and the data dimension are positive and significant for most items, suggesting that teachers who use AI more frequently also tend to use data more competently in their educational practice.

The only exception to these correlations was the item “Knowing the organizational structure, roles, and internal processes of the educational institution related to the handling and dissemination of data in the educational context of the classroom”, which did not show a significant relationship with AI use. This may indicate that institutional knowledge about data management is not directly related to the use of AI in pedagogical tasks.

The data-handling items that showed the highest correlations with AI use include

• “Using data from digital learning environments to identify learning problems of students and the groups I teach”;
• “Using data from digital environments to solve problems in educational practice, both with individual students and as a group”;
• “Recording the solutions to educational problems identified through data using reports, memos, diaries, or other tools”;
• “Using evaluation, content transmission, or visualization tools that allow working with raw data to obtain automatic and detailed results”;
• “Modifying the design of learning activities according to problems detected through data generated in digital environments”;
• “Considering data on student performance and learning in digital environments when implementing learning activities”;
• “Taking into account data on student activity in digitized environments when evaluating teaching activities”, with a correlation of 0.28 (p < 0.01), highlighting this last item as having the highest significant correlation.

4. Discussion

This research explores the relationships among AI knowledge, use, perceived risks, and data literacy among teachers, addressing the stated research objectives (O1, O2, and O3). Descriptive statistics offer a context for interpreting these relationships. As noted, the sample comprises a slight majority of women (60.8%), suggesting a potential gender dynamic in AI adoption. The concentration of AI users among teachers aged 36–60 hints at the influences of experience and professional maturity on technology integration. The prominence of STEM (35.8%) and language (29.8%) teachers in AI adoption aligns with the perceived suitability of AI for data-driven and analytical disciplines. The prevalence of AI use among fourth year ESO teachers (42.3%) and the high percentages of full-time teachers (89.7%) and those with over 16 years of experience (47.5%) provide further demographic context for understanding AI adoption patterns (O1).

The observed dispersion in responses regarding advanced AI use reveals significant variability in implementation and familiarity with these tools, indicating areas of underutilization. This range of adoption, with some teachers demonstrating advanced usage while others remain at low or moderate levels, pinpoints specific AI tools and applications that are currently underutilized in teaching practice. For example, although text generation and content detection are used by 17.1% and 12.3% of the teachers frequently or daily, respectively, video generation and simulations see much lower usage (2.4%), highlighting a gap between potential and practice usages. This dispersion likely reflects contextual influences, such as resource availability and institutional policies, impacting both the frequency and nature of AI adoption and further informing O1.

The correlations between AI use for specific tasks (e.g., code and image generation) and concrete pedagogical objectives (e.g., lesson planning and material creation), with 22% of the teachers using AI for lesson planning often or constantly and 17.1% for material creation, directly address this issue by demonstrating a clear link between AI use and pedagogical goals. This targeted approach suggests that teachers are strategically leveraging AI to enhance their existing pedagogical practices, specifically in the design and creation of educational resources. Furthermore, the exploration of the correlations among AI knowledge, use, and perceived risks, reveals a complex interplay. The positive correlations between AI knowledge (53.7% reporting moderate or high knowledge) and the frequency of use point toward a bidirectional relationship, suggesting that knowledge facilitates use, and use, in turn, enhances knowledge. Conversely, the negative correlation between the frequency of AI use and concerns about its classroom effects further addresses this issue by showing that experience mitigates perceived risks, fostering a more balanced perspective. For instance, although ethical concerns and academic integrity are major concerns (24.4% extremely concerned and 58.5% very concerned), these concerns may lessen with increased usage (O2).

Finally, the positive correlations between data literacy and AI use, as discussed above, directly address this issue and highlight the critical link between these two competencies, demonstrating that teachers proficient in data handling and analysis are more likely to effectively integrate AI into their educational practice. This suggests that data literacy is fundamental for leveraging AI applications in education. However, the nuanced finding regarding the specific type of data literacy relevant to teaching, namely, practical classroom data handling rather than a broader understanding of organizational data structures, provides a more granular understanding of the relationship between AI use and data literacy, fulfilling the objective O3. For example, although many teachers feel capable of basic data tasks, fewer feel confident in more advanced skills, like monitoring student performance through digital learning environment data (14.7% are not capable). In summary, the discussion presented herein, enriched by the descriptive data, directly addresses each of the stated research objectives, providing valuable insights into the current state of AI integration in education (O3).

5. Conclusions

This study reveals a nuanced landscape of AI awareness and adoption among teachers in Catalan rural schools. Although a moderate level of awareness exists, practical classroom application remains limited, primarily focused on text generation and content detection. A gap between the theoretical understanding and practical application of data literacy skills is also evident. Key perceptual barriers to broader AI integration include ethical concerns, potential impacts on academic integrity, and the perceived risk of diminishing critical thinking skills. However, these findings indicate a positive correlation between AI knowledge and its diverse range of applications, suggesting that familiarity breeds adoption. Higher AI usage correlates with reduced concern about potential risks, implying that direct experience fosters a more balanced perspective. Furthermore, the observed link between data literacy and AI use underscores the crucial role of data handling and analysis skills in effectively leveraging AI within education. This aligns with existing research (Chen & Chang, 2024; Heeg & Avraamidou, 2023; González-Calatayud et al., 2021) highlighting AI’s potential for personalized learning, targeted assistance, and improved teaching methodologies. Specifically, AI algorithms can analyze student data to inform early interventions and customize learning trajectories, thereby enhancing learning outcomes by adapting educational experiences to individual needs.

These findings necessitate enhanced training and support programs to facilitate effective AI integration. Such programs should prioritize developing competencies in both AI tool utilization and data handling, ultimately contributing to improved learning outcomes and promoting educational equity within rural communities. This could involve integrating personalized learning platforms, intelligent tutoring systems, and automated administrative tools. Future research should explore the potential of AI-driven personalized learning platforms to address individual learning gaps and tailor educational content to specific students’ needs within the rural context. Additionally, investigating the efficacy of AI-powered assessment tools in providing timely and actionable feedback to both teachers and students is warranted.

A limitation of this study is the observed dispersion in responses, reflecting the heterogeneity in teachers’ perceptions and knowledge, which limits broader generalizations. Future studies could employ larger and more representative samples to address this limitation. Critically, AI integration must address broader challenges within rural education, including unequal resource distributions, teacher shortages, and deficits in professional expertise. Future research should investigate how AI can be strategically deployed to mitigate these existing structural inequalities. Specifically, exploring the role of AI in providing remote access to specialized resources and expertise in underserved rural areas is crucial. Addressing teacher attitudes and anxieties through targeted professional development is essential for successful AI integration. Future interventions should focus on building teacher confidence and competence in using AI tools effectively. By strategically deploying AI and addressing these multifaceted challenges, we can maximize its potential to enhance learning outcomes and promote educational equity in rural communities. This includes exploring the development of AI-powered tools specifically designed to support teachers in rural schools, such as automated grading systems or personalized learning content creation tools. Further research should investigate the long-term impacts of AI integration on students’ learning outcomes and teachers’ professional development in rural settings.