Artificial Intelligence in Mathematics Education: Preservice Teachers’ Perceptions of Expectations, Practices, and Challenges*
Robinson Junior Conde-Carmona
Universidad del Atlántico, Barranquilla (Colombia)
https://orcid.org/0000-0002-7421-1754
Iván Andrés Padilla-Escorcia
Universidad del Atlántico, Barranquilla (Colombia)
https://orcid.org/0000-0003-1210-3712
Reception: October 7, 2024 | Acceptance: November 19, 2024 | Publication: May 31, 2025
DOI: http://doi.org/10.18175/VyS16.2.2025.1
ABSTRACT
This study explores how future math teachers perceive integrating artificial intelligence (AI) into their professional practice. Using a phenomenological-hermeneutic approach, the expectations, anticipated practices, and perceived challenges of thirty preservice teachers from a public university in Barranquilla, Colombia, were examined. Data was gathered through semi-structured interviews, focus groups, reflective journals, and field observations. Thematic analysis showed enthusiasm for AI’s potential to personalize learning and make abstract concepts more visual, along with concerns about equity, privacy, and reliance on technology. Participants viewed AI as a “copilot” in teaching and stressed the importance of comprehensive training to use this tool effectively. The findings highlight the need for a balanced teacher training approach that considers both the opportunities and challenges AI presents in mathematics education.
KEYWORDS
artificial intelligence, educational technology, mathematics education, teacher perceptions, teacher training.
La inteligencia artificial en educación matemática: percepciones de futuros docentes sobre expectativas, prácticas y desafíos
RESUMEN
Este estudio examina las percepciones de futuros docentes de matemáticas sobre la integración de la inteligencia artificial (IA) en su práctica profesional. Utilizando un diseño fenomenológico-hermenéutico se investigaron las expectativas, prácticas anticipadas y desafíos percibidos de treinta profesores en formación de una universidad pública en Barranquilla, Colombia. Se recolectaron datos mediante entrevistas semiestructuradas, grupos focales, diarios reflexivos y observaciones de campo. El análisis temático reveló un entusiasmo por el potencial de la IA para personalizar el aprendizaje y visualizar conceptos abstractos, al igual que evidenció preocupaciones sobre equidad, privacidad y dependencia tecnológica. Los participantes describieron la IA como un “copiloto” en la enseñanza y enfatizaron en la urgencia de una capacitación integral en el uso de esta herramienta. Los resultados sugieren la importancia de adoptar un enfoque equilibrado en la formación docente que aborde tanto las oportunidades como los desafíos de la IA en la educación matemática.
PALABRAS CLAVE
educación matemática, formación docente, inteligencia artificial, percepciones docentes, tecnología educativa.
Inteligência artificial na educação matemática: percepções de futuros professores sobre expectativas, práticas e desafios
RESUMO
Este estudo examina as percepções de futuros professores de matemática sobre a integração da inteligência artificial (IA) em sua prática profissional. Utilizando um design fenomenológico-hermenêutico, foram investigadas as expectativas, práticas antecipadas e desafios percebidos. Participaram 30 professores em formação de uma universidade pública em Barranquilla, Colômbia. Os dados foram coletados por meio de entrevistas semiestruturadas, grupos focais, diários reflexivos e observações de campo. A análise temática revelou entusiasmo pelo potencial da IA para personalizar a aprendizagem e visualizar conceitos abstratos, juntamente com preocupações sobre equidade, privacidade e dependência tecnológica. Os participantes visualizaram a IA como um “copiloto” no ensino, enfatizando a necessidade de uma formação abrangente em IA. Os resultados sugerem a necessidade de uma abordagem equilibrada na formação de professores que aborde tanto as oportunidades quanto os desafios da IA na educação matemática.
PALAVRAS-CHAVE
educação matemática, formação de professores, tecnologia educacional, percepções de professores, inteligência artificial.
Introduction
The integration of artificial intelligence (AI) into education marks a major paradigmatic shift that is changing traditional pedagogical practices. Especially in mathematics education, this technological advancement offers unique opportunities to address various challenges in teaching and learning. This study explores how preservice mathematics teachers perceive this change by analyzing their expectations, experiences, and concerns about integrating AI into their professional practice.
AI is rapidly transforming various parts of society, and education is no different. Recently, there has been growing interest in using AI technologies to improve student learning outcomes (Baker et al., 2019; Chen, Chen et al., 2020; Hwang et al., 2020). This phenomenon has led researchers and educators to see AI as a potential solution to ongoing challenges in education, like the shortage of qualified teachers and the need for more efficient and personalized educational resources (Cope et al., 2020; Guan et al., 2020).
In the field of mathematics education, AI presents particularly promising opportunities. Often considered a “gatekeeper course” (Harper et al., 2021), mathematics is crucial for students’ academic success, future careers, and overall social development. However, many students still face difficulties in this subject, emphasizing the need for innovative teaching and learning methods.
The integration of AI in mathematics education spans a wide range of applications, including intelligent tutoring systems, automated assessment tools, and adaptive learning platforms (Gulz et al., 2020; Hasanein & Abu-Naser, 2018). These technologies have the potential to provide personalized learning experiences, deliver immediate feedback, and foster a deeper and more meaningful understanding of mathematical concepts (Panqueban & Huincahue, 2024).
However, despite the excitement surrounding AI in education, its successful implementation mainly depends on teachers’ attitudes, perceptions, and practices when using it in the classroom (Duzhin & Gustafsson, 2018). Mathematics teachers, in particular, are in a unique position to influence how AI tools are adopted and how effective they are in teaching this important subject.
Interestingly, research on mathematics teachers’ perspectives on using AI in their teaching practices is relatively scarce (Chen, Xie et al., 2020). This gap in the literature is particularly important because teachers are crucial in successfully adopting new educational technology. Understanding their expectations, practices, and challenges related to AI is vital for effectively and positively incorporating these tools into the classroom.
In this context, the current study aims to address this knowledge gap by exploring the following research questions:
Studies show that traditional mathematics teaching methods are becoming less effective at motivating students. Chen, Xie et al. (2020), Hwang et al. (2020), and Fang et al. (2019) note that approaches like recitation, dialogue, and discussion no longer inspire enthusiasm and motivation in learning math. This growing gap between traditional methodologies and current educational needs highlights the urgent need to adopt more innovative strategies, especially those based on AI. Integrating AI systems and applications into mathematics education can create new opportunities to increase student interest and improve teaching effectiveness, addressing the need to update pedagogical approaches in this field. This study aims to explore mathematics teachers’ perceptions, practices, and challenges regarding AI to inform education policies, professional training programs, and future research, ultimately contributing to a more effective adoption of AI in mathematics teaching.
This study not only has implications for immediate educational practice but also supports broader sustainable development goals in education, such as those outlined by the United Nations Educational, Scientific and Cultural Organization (UNESCO, 2019). By examining how AI can foster a more inclusive, equitable, and high-quality mathematics education, this research aims to contribute to the progress of education in the digital age and the development of essential mathematical skills.
Theoretical Framework
Procedural Definitions: AI in Education
AI in the educational setting refers to the technology, software, methodologies, and computer algorithms used to solve problems related to human learning (Chen, Chen et al., 2020). It encompasses a wide range of technological tools—from software and hardware to educational robots and mobile apps—designed to interpret patterns in collected data (such as students’ understanding and mistakes) and to make informed decisions that suggest next steps to improve learning outcomes (Hwang et al., 2020). This conceptual framework also recognizes the challenges that mathematics teachers face when implementing these technologies, including obstacles that may prevent their effective use in teaching.
Among the different types of AI, the most commonly used in education are intelligent tutoring systems (ITS), adaptive learning systems (ALS), and robotics (Chu et al., 2021). These have been employed to enhance teaching and learning outcomes in mathematics (Bush, 2021; Fanchamps et al., 2021; Tashtoush, 2019).
Teachers’ Perceptions of the Use of AI
AI has not been fully adopted in classrooms because many teachers still hold negative attitudes toward technology and choose not to use it (Nguyen, 2023; Rasheed & Tashtoush, 2023; Shirawia et al., 2023). Teachers’ anxiety about using new tools and a preference for staying within their comfort zone are some factors that hinder the integration of technology in face-to-face teaching (Kim, 2023).
However, recent research has led to an increase in teachers’ expectations in this area, aiming for significant changes in education, such as the use of AI in various educational settings (Méndez Parra & Conde-Carmona, 2025). Teachers’ perceptions of AI in education (AIED) vary based on their pedagogical beliefs, teaching experiences, prior familiarity with educational technology, and the effectiveness and necessity of specific tools (Wang et al., 2020).
Studies have examined how university and elementary school teachers use AI applications, with results generally showing low levels of usage (Shin & Shin, 2020; Wang et al., 2020). Other research has explored students’ attitudes toward AI ethics and how this technology influences elementary school students’ mathematics achievement (Hwang, 2022; Jang et al., 2022).
Additional research has employed the Technology Acceptance Model (TAM) to explore teachers’ perceptions of factors affecting the adoption of AI applications in science education (Mahmoud, 2020). Positive correlations have been found with self-efficacy, ease of use, anticipated benefits, attitudes, and behavioral intentions.
In conclusion, the integration of AI into mathematics education continues to advance rapidly, bringing new opportunities and challenges. Understanding mathematics teachers’ perceptions, practices, and challenges related to AI is crucial for a successful and beneficial implementation of these technologies in the classroom.
Methodology
Study Design
The present study adopted a qualitative approach using a phenomenological-hermeneutic research design (Van Manen, 2016). This method allowed us not only to explore preservice teachers’ lived experiences with using AI in mathematics instruction but also to interpret and situate these experiences within the larger context of mathematics education and technology integration. The selection of this design is based on its ability to capture the core of perceptions, expectations, and challenges encountered by preservice teachers, while offering an in-depth interpretation of the underlying meanings (Creswell & Poth, 2018).
Participants and Context
The study included thirty preservice teachers (eighteen women and twelve men) from five professional practice groups in Mathematics Education 1 and 2 at a public university in Barranquilla, Colombia. Participants ranged in age from 20 to 25 years old and represented diverse socioeconomic levels, with 60% from strata 1-2, 30% from strata 3-4, and 10% from strata 5-6. Maximum variation purposive sampling (Patton, 2015) was used to ensure a wide range of perspectives.
The selection criteria included prior experience with educational technology (at least one year), years of teacher training (between four and five years), and area of specialization within mathematics. This selection provided a diverse and comprehensive view of the experiences of future teachers at different stages of their practical training and with varying levels of exposure to educational technology.
Data Collection Techniques and Instruments
To ensure a comprehensive understanding of the phenomenon studied, a methodological triangulation strategy (Denzin, 2017) was employed using multiple data collection methods during an academic semester in the period 2024-2021.
Data Analysis
Before analysis, the researchers transcribed all audio recordings from interviews, focus groups, and field observations word-for-word. The reflective journals and field notes were digitized and added to a structured database along with the transcripts. This process of data organization and preparation happened alongside data collection, enabling a preliminary analysis that helped inform later data collection phases.
Data analysis was conducted using a reflexive thematic analysis approach (Braun & Clarke, 2019), which acknowledges the active role of the researcher in interpreting the data. This iterative process included the following phases:
Throughout the analysis process, analytical notes were recorded to document the researchers’ decisions and reflections. Additionally, the research team held regular meetings to discuss emerging themes and address any discrepancies in data interpretation.
Rigor and Reliability
To guarantee the rigor and reliability of the study, the strategies outlined in Table 1 were implemented.
Table 1. Strategies used to ensure the rigor and reliability of the study
|
Strategy |
Description |
|
Triangulation |
Triangulation of data, methods, and researchers was employed to enhance the reliability of the findings. |
|
Participant verification |
Emerging themes and initial interpretations were shared with participants for validation. |
|
External audit |
An independent researcher reviewed the analysis process and results to ensure they matched and were backed by the data. |
|
Reflexivity |
Researchers kept reflective journals during the research process to document and evaluate their own assumptions and biases. |
Source: Own elaboration.
Ethical Considerations
Informed consent was obtained from all participants, and they were assured of confidentiality and their right to withdraw from the study at any time without consequences.
Discrepant Data Handling
Special attention was paid to discrepant data or outliers in the analyses, recognizing their potential to deepen our understanding of the studied phenomenon. These were identified as data points that significantly deviated from the majority view. For example, while most participants expressed optimism about AI’s potential, three participants showed notable skepticism. These cases were carefully examined to understand the reasons behind this difference. Discrepant data were included in our overall thematic analysis and assigned a dedicated subsection in the Results section titled “Diverging Perspectives.” Addressing discrepancies in this way allowed us to capture the diversity of viewpoints among preservice mathematics teachers and provide a more comprehensive and nuanced portrayal of their perceptions of AI in education.
Results
Transformative Potential of AI in Mathematics Teaching
Across various data sources, this central theme emerged as a common thread, highlighting future teachers’ optimism and expectations about how AI could transform the teaching and learning of mathematics.
Personalization of learning
AI’s ability to deliver personalized learning experiences was a prominent subtheme, mentioned by 28 of the 30 participants. This frequent mention shows a strong consensus on AI’s potential to address individual student needs in ways traditional methods cannot.
As a participant eloquently stated,
AI can analyze each student’s progress and customize math exercises to their individual level of understanding. This is something I, as a teacher, couldn’t do manually for 30 students at once. (Participant 12, interview)
This perspective aligns with existing research on the benefits of adaptive instruction in mathematics (Mousavinasab et al., 2021). Participants envisioned scenarios where AI systems could perform these tasks:
However, some participants also voiced caution:
Although personalization appears helpful, I’m worried that it could lead to a more isolated learning experience. Math also benefits from peer interaction. (Participant 19, Focus Group 3)
The above aligns with the findings from the two observation units involving Participant 19. This preservice teacher encourages collaborative work as a way for students to understand mathematical concepts through peer feedback, focusing on identifying potential errors in solving classroom problems. This highlights the need to balance personalization with collaborative learning opportunities, a challenge that future advancements in educational AI will need to address.
Visualization of abstract concepts
The ability of AI to create dynamic and interactive visualizations of abstract mathematical concepts was highlighted by 25 of the 30 participants. Its potential to make complex ideas more accessible was appreciated. One participant shared: “Imagine being able to show students how a function transforms in real time by adjusting parameters with voice commands. AI could make this possible” (Participant 3, focus group 2).
Participants suggested scenarios such as manipulable 3D simulations, interactive visualizations of statistical concepts, and explorable graphical representations of optimization problems, which align with research on visualization technologies in mathematics education (Lovett & Knezek, 2023). However, a warning was raised regarding proper use: “We need to ensure that AI visualizations do not become a crutch. Students still need to develop the ability to visualize concepts mentally” (Participant 8, reflective journal). This observation emphasizes the importance of integrating AI tools in a way that supports the development of fundamental cognitive skills.
Continuous assessment and immediate feedback
AI’s ability to provide continuous assessment and immediate feedback emerged as a key subtheme, highlighted by 27 of the 30 participants. They valued this feature for its potential to transform assessment from a one-time event into an ongoing, formative process. One participant reflected in their journal: “With AI, we could obtain detailed analyses of each student’s problem-solving process, not just the final answer. This would allow us to intervene more effectively” (Participant 7, reflective journal).
Participants outlined various potential uses:
These ideas align with research on the positive effects of immediate feedback in mathematics learning (Kiili & Tuomi, 2019). However, some participants also expressed concerns: “I am concerned that overreliance on AI feedback may diminish students’ ability to self-assess and develop metacognition” (Participant 16, interview).
Regarding the question about AI’s role in assessment and feedback in mathematics teaching, participants agreed that this technology is a valuable tool for formative assessment. Its ability to display instant results allows for dynamic adjustment of the difficulty level of mathematical problems and supports personalized rather than collective feedback. These features make it possible to analyze both meaningful learning and the specific difficulties faced by each student, especially those related to mathematical concepts not yet solidified in previous years. The findings emphasize the importance of developing AI systems that, in addition to providing feedback, promote the development of students’ metacognitive skills.
Anticipated Challenges in Implementing AI
While the transformative potential of AI generated enthusiasm among participants, a set of anticipated concerns and challenges also emerged. This theme reflects a mature awareness of the complexities involved in integrating AI into mathematics education.
Digital divide and unequal access
Concern about the digital divide and unequal access to AI technologies was a major subtheme, mentioned by 22 of the 30 participants. This frequent mention shows a strong awareness of the ethical and equity issues related to implementing AI in education. One participant clearly expressed this concern: “I am concerned that implementing AI could exacerbate existing inequalities. Not all students have access to devices or the internet at home” (Participant 1, focus group 4).
This precaution was shown in various aspects.
Unequal access to hardware: Concerns about students lacking the proper devices to run advanced AI applications.
The above points align with the literature on the digital divide in education (Wang et al., 2020) and highlight the importance of addressing these inequalities as a central part of any educational AI initiative. Some participants proposed potential solutions.
We need to consider hybrid models that combine AI with traditional methods, ensuring no student is left behind due to limited access to technology. (Participant 25, interview)
The only way to ensure that AI use doesn’t create gaps in learning an exact science like mathematics is for schools to have digital technologies that are both quantitative and qualitative. A student with access to these tools and knowing how to use them can develop critical, creative, and logical-mathematical thinking to solve problems in this area from different perspectives. (Participant 4, interview)
It is worth noting that the perspectives of Participants 4 and 25 align with those of Padilla Escorcia and Conde-Carmona (2020) in that, regardless of the modality (in-person, virtual, or hybrid), the use of digital technologies helps facilitate mathematics teaching. This emphasizes a pragmatic approach that acknowledges current educational realities while working toward a more equitable future.
Cognitive overload and reliance on technology
Another key subtheme was concern about potential cognitive overload and overreliance on technology, mentioned by 18 of the 30 participants. Although less frequent than other topics, discussions on this subtheme were particularly intense and reflective. One participant expressed a similar concern: “I fear that students may become overly reliant on AI tools and lose fundamental math skills” (Participant 23, interview).
This concern manifested in several forms:
These concerns are supported by research on the potentially negative effects of technology on learning (Panqueban & Huincahue, 2024), highlighting the need for a balanced approach to AI integration. Some participants suggested ways to reduce these risks: “We must design activities that combine the use of AI with traditional methods, fostering a balance between technological assistance and the development of independent cognitive skills” (Participant 10, focus group 5). This statement demonstrates a pedagogical approach that aims to leverage AI’s benefits while also fostering key cognitive skills.
Data privacy and security
Concerns about student data privacy and security also emerged as a key subtheme, mentioned by 24 of the 30 participants—a high frequency that indicates a strong awareness of the ethical and legal issues related to using AI in education. In this context, one participant asked an important question: “How can we ensure that our students’ personal data and academic performance are protected when using AI systems?” (Participant 5, Focus Group 1).
This concern was broken down as follows:
This aligns with current debates on AI ethics in education (Hwang et al., 2020) and highlights the need for strong data governance frameworks in implementing educational AI. Some participants suggested possible measures to address this issue: “We need to develop clear and transparent policies on the use of data in educational AI systems and educate students and parents about their rights and the safeguards in place” (Participant 29, reflective journal). This is a proactive approach that seeks to address privacy and security concerns through transparency and education.
Overall, these results offer a detailed and nuanced understanding of preservice mathematics teachers’ perceptions of using AI in their professional practice. They demonstrate both enthusiasm for AI’s transformative potential and a clear awareness of the challenges and ethical concerns involved. These findings highlight the need for a careful and balanced approach to integrating AI into mathematics education—one that aims to maximize its advantages while actively addressing potential challenges.
Preparation and Vision for a Collaborative Future Between Teachers and AI
Data analysis revealed a common theme connecting the need for targeted AI training for educators with the vision of a collaborative future between teachers and technology. This theme highlights the participants’ awareness of the importance of preparing for a time when AI becomes an integral part of mathematics education, along with their ideas of what that future might look like.
Development of comprehensive skills in educational AI
Participants expressed a wide range of needs to develop skills, from technical abilities to pedagogical and critical thinking skills. This subtheme was mentioned by 29 out of 30 participants, demonstrating near-unanimous agreement on its importance. One participant articulated it well: “We need practical training in the use of AI tools. Not just theory, but hands-on experiences with real-world applications. But beyond that, we need to learn how to integrate these tools in pedagogically meaningful ways and how to critically evaluate them” (Participant 15, reflective journal).
This comprehensive view included the following dimensions:
These concerns relate to recent research on teacher preparation for the AI era (Panqueban & Huincahue, 2024), which emphasizes the need for a multidimensional approach to teacher training. One participant briefly summarized the importance of critical evaluation: “We need to know how to assess the quality and reliability of AI applications before using them in our classrooms. We can’t simply assume that because it’s AI, it’s good or appropriate” (Participant 4, focus group 5). This comment demonstrates a mature understanding that AI, like any educational tool, must be carefully evaluated for its pedagogical value and fit within the specific learning environment.
Redefining the role of teachers in the AI era
As participants reflected on their training needs, they also shared their vision for how AI could transform the role of teachers in the future. This subtheme was mentioned by 27 of the 30 participants, highlighting a widespread consideration of how AI might reshape their future professional practices. One participant shared an inspiring insight:
I view AI as a copilot on my teaching journey, helping me navigate challenges and strengthen my skills as an educator. It won’t replace my role but will allow me to be more effective and focus on the truly human parts of teaching. (Participant 28, interview)
This idea of AI as a “copilot” or teacher assistant has shown up in various ways.
The teachers’ expectations align with the literature on the future of teaching in the AI era (Kiili & Tuomi, 2019), which suggests an evolution of the teacher’s role toward becoming a facilitator and designer of technology-enriched learning experiences. As one participant noted in their journal:
With AI taking care of tasks like grading and data analysis, I will be able to devote more time to mentoring and guiding my students. I envision my future role as a learning experience designer and as a facilitator of curiosity and critical thinking. (Participant 6, reflective journal)
This reflection encourages us to view the teacher as a guide who uses AI to create richer and more personalized learning environments, rather than being replaced by AI.
Toward interdisciplinary collaboration in mathematics education with AI
A final subtheme was the recognition of the need for interdisciplinary collaboration to fully realize AI’s potential in mathematics education, mentioned by 20 of the 30 participants, reflecting a growing awareness of the multifaceted nature of integrating AI into education. One participant described this need as follows: “To truly realize the potential of AI in mathematics education, we will need to collaborate closely with AI experts and educational technology designers. We cannot do this alone as educators; we need a truly interdisciplinary approach” (Participant 2, focus group 3).
This vision of interdisciplinary collaboration included the following variants:
Recent literature on the development of educational technologies, emphasizing co-design approaches and multidisciplinary teamwork (Hwang et al., 2020), aligns with the points discussed above. One participant remarked on the potential of collaboration:
I envision a future where interdisciplinary teams of educators, AI experts, designers, and researchers work together to develop intelligent learning ecosystems that genuinely transform mathematics education. It’s exciting to think about the opportunities this could create. (Participant 22, interview)
This testimony demonstrates a sophisticated understanding that the successful integration of AI into mathematics education requires the coming together of multiple perspectives and areas of expertise.
The findings above have important implications for developing teacher training programs, shaping educational policies, and guiding future research at the intersection of AI and mathematics education. They stress the need for a holistic and interdisciplinary approach to prepare future teachers for a rapidly changing educational environment, where AI will play an increasingly central role.
Results Table
Table 2. Frequency of subthemes in participants’ responses
|
Central theme |
Subtheme |
Frequency |
Mean |
Standard deviation |
|
Transformative potential |
Personalization of learning |
0.93 |
0.88 |
0.05 |
|
Visualization of abstract concepts |
0.83 |
|||
|
Continuous assessment and immediate feedback |
0.9 |
|||
|
Anticipated challenges |
Digital divide and unequal access |
0.73 |
0.7 |
0.1414 |
|
Cognitive overload and reliance on technology |
0.6 |
|||
|
Data privacy and security |
0.8 |
|||
|
Training needs |
Technological competencies |
0.96 |
0.86 |
0.1 |
|
Pedagogical integration of AI |
0.86 |
|||
|
Critical evaluation of tools |
0.76 |
|||
|
Collaborative vision |
AI as a teacher’s assistant |
0.9 |
0.8 |
0.120 |
|
Redefinition of the teaching role |
0.83 |
|||
|
Interdisciplinary collaboration |
0.66 |
Source: Own elaboration.
Table 2 shows the participants’ overall positive attitude toward AI in mathematics education. Concerns about student dependency, indicated by anticipated and cognitive overload, had the lowest mean score among the subthemes (0.6). Regarding the main themes, anticipated challenges had the lowest mean score (0.7), followed by collaborative vision (0.8), the need for training (0.86), and transformative potential (0.88).
Participants view AI as a transformative strategy but approach it cautiously, stressing it should not dominate education. The transformative potential had the lowest standard deviation (0.05), showing agreement on its benefits in personalization, visualization, and continuous assessment. In contrast, the anticipated challenges had greater dispersion (0.1414), reflecting concerns about educational oversimplification that could downplay the effort required to learn mathematics.
Diverging Perspectives
Most participants expressed positive views on integrating AI into mathematics education and highlighted its potential to personalize learning. They especially valued AI’s ability to adjust exercises and problems to accommodate different learning speeds, including students with special educational needs and gifted students. Participants also appreciated how AI could provide feedback to multiple students simultaneously and enable a deeper analysis of their progress.
However, some divergent voices arose, expressing concerns. Certain participants, while acknowledging the benefits of AI, prefer to limit its use to avoid potential student addiction to these tools. There is worry that excessive AI use could interfere with the study of mathematical formalism. These minority perspectives offer valuable insights into the reservations some preservice mathematics teachers have about integrating AI into education.
Because of this, and based on the perceptions of preservice teachers who do not consistently use AI in their classes, the focus groups were asked the following question: How do you think AI could help visualize abstract mathematical concepts? Most participants said that these tools offer a variety of options for students learning abstract content, such as proofs in analytical and Euclidean geometry, as well as graphically calculating the domain and range of functions. This echoes the testimony of Participant 17:
Artificial intelligence allows students to manipulate and explore mathematical representations in real time. It can also create a variety of illustrative examples and problems that show how abstract concepts are applied in different contexts, helping students better understand the applications and relevance of what they are learning. (Participant 17, interview)
This perspective from Participant 17 aligns with what was observed during two of his class sessions. The preservice teacher is completing his professional internship in Mathematics Education 2 at a public school in Barranquilla, where there are no technological resources to support teaching in this subject. However, he is resourceful and uses one of the school’s three video projectors to show students, using an AI called Mathway, the detailed process of how to calculate the domain and range of a rational function. Afterward, he encourages students to try to explain the step-by-step process of calculating the domain and range of a rational function based on the process provided by the AI.
Skepticism about the pedagogical value of AI
Three participants (out of 30 in the sample) expressed serious skepticism about the true pedagogical value of AI in teaching mathematics. They argued that a deep understanding of mathematics requires reasoning that AI cannot yet replicate or effectively support. One of them commented:
I worry that AI might give quick answers without encouraging deep understanding. Mathematics isn’t just about getting the right solution; it’s about understanding the process. I’m not convinced that AI can develop that kind of thinking. (Participant 24, interview)
Concern over the dehumanization of education
Two participants voiced strong concern about the potential dehumanization of mathematics education due to widespread AI use. For them, the human connection in teaching is irreplaceable, and overreliance on AI could weaken vital aspects of the teacher-student relationship: “Teaching mathematics is not just about transmitting knowledge, but also about inspiring, motivating, and connecting with students. I fear that by relying too heavily on AI, we will lose that crucial human element” (Participant 8, focus group 3).
Questions about accessibility and equity
Although most participants saw AI as a tool to enhance accessibility, four participants expressed concerns that AI could make existing inequalities in mathematics education worse: “I am concerned that AI could disproportionately benefit schools and students who already have resources. What about schools in low-income or rural areas? We could be creating a new form of digital divide” (Participant 17, reflective journal).
Resistance to pedagogical change
Two participants expressed a preference for traditional teaching methods and were hesitant to adopt AI in their future practice: “I have experienced the power of traditional mathematics teaching, and I am not convinced that we need to reinvent the wheel. Sometimes, a good teacher with chalk and a blackboard can achieve more than any advanced technology” (Participant 5, interview).
These diverging perspectives reveal the complexity of integrating AI into mathematics education and the wide array of opinions among future teachers. They highlight the importance of a balanced approach to AI adoption that considers these concerns and actively addresses them in teacher training and education policy development.
Additional Findings
These findings provide a detailed and nuanced understanding of preservice mathematics teachers’ perceptions, expectations, and concerns about incorporating AI into their professional practice. They emphasize the importance of a balanced approach to teacher training that considers both the opportunities and challenges AI brings to mathematics education.
Discussion
The findings of this research directly address our initial questions and reveal a complex and nuanced view of preservice teachers’ perceptions of AI in mathematics education. Regarding expectations, a large majority of participants (93%) foresee significant benefits in personalizing learning, while 83% especially value the potential for visualizing abstract mathematical concepts, and 90% anticipate notable improvements in continuous assessment processes. Concerning the current integration of AI, emerging practices in visualizing complex mathematical ideas were identified, with a particular focus on its use for formative assessment and initial testing of automated feedback tools. Nevertheless, the identified challenges are also notable: 73% of participants voiced concerns about equity in accessing these technologies, 60% worried about potential overreliance on technology, and 80% emphasized the importance of strong protocols to protect student data. Overall, these results show that although there is a lot of optimism about AI’s transformative potential in mathematics education, preservice teachers remain critical and thoughtful about how it is implemented.
Thus, this qualitative study reveals a complex view of preservice mathematics teachers’ expectations and concerns about integrating AI into their professional practice. Participants expressed strong enthusiasm for the opportunities this technology provides for personalizing learning and visualizing abstract concepts, which aligns with the findings of Chen, Chen et al. (2020) about AI’s transformative potential in education. However, their optimism is tempered by significant worries about equity in access and ethical issues, echoing warnings from Hwang et al. (2020) regarding the need to address gaps in the use and understanding of educational AI.
Participants’ perception of AI as a “copilot” in teaching aligns with Wang et al. (2020) regarding teachers’ willingness to adopt intelligent tutoring systems. However, our findings reveal a more nuanced understanding of how AI could reshape the teaching role. Beyond simple technological adoption, this signifies a deep rethinking of pedagogical practice, consistent with the observations of Guan et al. (2020) on the historical development of AI innovation in education.
The expressed need for comprehensive AI training for educators, covering technical, pedagogical, and ethical skills, aligns with Kim’s (2023) recommendations on AI-enabled curriculum design for primary and secondary education. A comprehensive strategy is proposed, including AI fundamentals, its educational applications, and ethical considerations. However, based on the study’s findings, preservice teachers anticipate a deeper and more transformative integration of AI, going beyond merely introducing tools and suggesting a fundamental redefinition of teaching practices in an educational setting increasingly mediated by smart technologies.
Participants’ concerns about the digital divide and privacy issues reflect the challenges identified by Méndez-Parra and Conde-Carmona (2025) in integrating AI into education. However, based on our results, preservice teachers do not see these challenges as insurmountable barriers but rather as important factors that need to be actively addressed in teacher training and educational policy development. This perspective offers an interesting contrast to the existing literature, which indicates that preservice teachers are more prepared and forward-thinking than previously reported.
The sample of thirty preservice teachers was selected based on the criteria of theoretical saturation in phenomenological research (Guest et al., 2006), which suggests that choosing between twenty and thirty participants provides enough depth for qualitative studies. According to Creswell and Poth (2018), this sample size enables a thorough exploration of individual experiences while keeping the data manageable for analysis.
Sample diversity was achieved through purposive sampling with maximum variation, including participants from different socioeconomic backgrounds, experience levels, and mathematical disciplines. This method improves the potential transferability of the findings to similar settings. However, the geographic scope is limited to a single institution in Barranquilla, Colombia, so caution is needed when generalizing the results.
As is common in qualitative research, data interpretation involves some subjectivity. To address this, we applied rigorous validation methods: methodological triangulation, participant verification, and external auditing. Potential influences of social desirability were managed through multiple data collection techniques.
We recognize that this study offers a snapshot in a rapidly evolving field like artificial intelligence in education, where perceptions shift quickly due to technological advances. Therefore, we included longitudinal monitoring throughout the academic semester to observe how participants’ perceptions change.
Finally, we acknowledge that our own perspectives as researchers can influence interpretation. To address this potential bias, we implemented a systematic reflexivity process, documented through analytical notes and regular team discussions, supported by a thorough peer review.
Conclusions
This study provides a unique insight into preservice mathematics teachers’ perspectives on integrating AI into their professional practice: it reveals a balance between enthusiasm for its transformative potential and a clear awareness of the challenges involved. Participants showed a sophisticated understanding of how AI could change mathematics teaching, especially in areas like personalizing learning and visualizing abstract concepts. However, they also expressed important concerns about access equity, data privacy, and the risk of becoming overly dependent on technology, emphasizing the need for an ethical and student-centered approach to AI implementation.
The emerging vision of a collaborative future between teachers and AI, where technology functions as a “copilot” that boosts the educator’s abilities, indicates an evolution in how the teacher’s role is viewed. This perspective, along with the acknowledgment of the need for comprehensive AI training covering technical, pedagogical, and ethical aspects, signals a paradigm shift in preparing future educators for a rapidly changing educational environment.
These findings have significant implications for designing teacher training programs, shaping educational policies, and guiding future research at the intersection of AI and mathematics education. They emphasize the need for a comprehensive and interdisciplinary approach that prepares future teachers not only to use AI tools but also to contribute to their development and ethical application. Future research should explore how to translate these perceptions and expectations into effective pedagogical practices and educational policies that support an equitable and pedagogically solid integration of AI into mathematics education.
References
Baker, T., Smith, L., & Anissa, N. (2019). Educ-AI-tion rebooted? Exploring the future of artificial intelligence in schools and colleges. Nesta. https://media.nesta.org.uk/documents/Future_of_AI_and_education_v5_WEB.pdf
Braun, V., & Clarke, V. (2019). Reflecting on reflexive thematic analysis. Qualitative Research in Sport, Exercise and Health, 11(4), 589–597. https://doi.org/10.1080/2159676X.2019.1628806
Bush, J. B. (2021). Software-based intervention with digital manipulatives to support student conceptual understandings of fractions. British Journal of Educational Technology, 52(6), 2299–2318. https://doi.org/10.1111/bjet.13139
Charmaz, K. C. (2014). Constructing grounded theory (2nd ed.). Sage Publications.
Chen, L., Chen, P., & Lin, Z. (2020). Artificial intelligence in education: A review. IEEE Access, 8, 75264–75278. https://doi.org/10.1109/ACCESS.2020.2988510
Chen, X., Xie, H., Zou, D., & Hwang, G.-J. (2020). Application and theory gaps during the rise of artificial intelligence in education. Computers and Education: Artificial Intelligence, 1, 100002. https://doi.org/10.1016/j.caeai.2020.100002
Chu, H.-C., Chen, J.-M., Kuo, F.-R., & Yang, S.-M. (2021). Development of an adaptive game-based diagnostic and remedial learning system based on the concept-effect model for improving learning achievements in mathematics. Educational Technology & Society, 24(4), 36–53. https://www.jstor.org/stable/48629243
Cope, B., Kalantzis, M., & Searsmith, D. (2020). Artificial intelligence for education: Knowledge and its assessment in AI-enabled learning ecologies. Educational Philosophy and Theory, 53(12), 1229–1245. https://doi.org/10.1080/00131857.2020.1728732
Creswell, J. W., & Poth, C. N. (2018). Qualitative inquiry and research design: Choosing among five approaches (4th ed.). Sage Publications.
Denzin, N. K. (2017). The research act: A theoretical introduction to sociological methods. Routledge. https://doi.org/10.4324/9781315134543
Duzhin, F., & Gustafsson, A. (2018). Machine learning-based app for self-evaluation of teacher-specific instructional style and tools. Education Sciences, 8(1), 7–21. https://doi.org/10.3390/educsci8010007
Fanchamps, N. L. J. A., Slangen, L., Hennissen, P., & Specht, M. (2021). The influence of SRA programming on algorithmic thinking and self-efficacy using Lego robotics in two types of instruction.International Journal of Technology and Design Education, 31, 203–222. https://doi.org/10.1007/s10798-019-09559-9
Fang, Y., Ren, Z., Hu, X., & Graesser, A. C. (2019). A meta-analysis of the effectiveness of ALEKS on learning. Educational Psychology, 39(10), 1278–1292. https://doi.org/10.1080/01443410.2018.1495829
Guan, C., Mou, J., & Jiang, Z. (2020). Artificial intelligence innovation in education: A twenty-year data-driven historical analysis. International Journal of Innovation Studies, 4(4), 134–147. https://doi.org/10.1016/j.ijis.2020.09.001
Gulz, A., Londos, L., & Haake, M. (2020). Preschoolers’ understanding of a teachable agent-based game in early mathematics as reflected in their gaze behaviors–An experimental study. International Journal of Artificial Intelligence in Education, 30(7), 38–73. https://doi.org/10.1007/s40593-020-00193-4
Harper, F., Stumbo, Z., & Kim, N. (2021). When robots invade the neighborhood: Learning to teach preK-5 mathematics leveraging both technology and community knowledge. Contemporary Issues in Technology and Teacher Education, 21(1), 19–52. https://citejournal.org/volume-21/issue-1-21/mathematics/when-robots-invade-the-neighborhood-learning-to-teach-prek-5-mathematics-leveraging-both-technology-and-community-knowledge
Hasanein, H. A. A., & Abu-Naser, S. S. (2018). Developing education in Israa University using intelligent tutoring system. International Journal of Academic Pedagogical Research, 2(5), 1–16.
Hwang, G.-J., Xie, H., Wah, B.-W., & Gašević, D. (2020). Vision, challenges, roles and research issues of artificial intelligence in education. Computers and Education: Artificial Intelligence, 1, 100001. https://doi.org/10.1016/j.caeai.2020.100001
Hwang, S. (2022). Examining the effects of artificial intelligence on elementary students’ mathematics achievement: A meta-analysis. Sustainability, 14(20), 13185. https://doi.org/10.3390/su142013185
Jang, Y., Choi, S., & Kim, H. (2022). Development and validation of an instrument to measure undergraduate students’ attitudes toward the ethics of artificial intelligence (AT-EAI) and analysis of its difference by gender and experience of AI education. Education and Information Technologies, 27(8), 11635–11667. https://doi.org/10.1007/s10639-022-11086-5
Kiili, K., & Tuomi, P. (2019). Teaching educational game design: Expanding the game design mindset with instructional aspects. In A. Liapis, G. N. Yannakakis, M. Gentile, M. Ninaus (eds.), Games and learning alliance: 8th International Conference, GALA 2019, Athens, Greece, November 27-29, 2019, proceedings (pp. 103–113). Springer International Publishing.
Kim, D. H. (2023). AI curriculum design for Korea K-12 AI education through analyzing AI education curriculum. International Journal of Recent Technology and Engineering, 12(3), 72–81. https://doi.org/10.35940/ijrte.C7173.0312323
Mahmoud, A. M. (2020). Artificial intelligence applications: An introduction to the development of education in light of the challenges of the corona virus (COVID-19) pandemic. International Journal of Research in Educational Sciences, 3(4), 171–224. https://www.iafh.net/index.php/IJRES/article/view/240
Méndez-Parra, C., & Conde-Carmona, R. J. (2025). Integración del enfoque STEAM y la realidad aumentada en la enseñanza de la traslación de figuras geométricas. Revista Virtual Universidad Católica del Norte, (74), 69–92. https://doi.org/10.35575/rvucn.n74a4
Mousavinasab, E., Zarifsanaiey, N., R. Niakan Kalhori, S., Rakhshan, M., Keikha, L., & Ghazi Saeedi, M. (2021). Intelligent tutoring systems: A systematic review of characteristics, applications, and evaluation methods. Interactive Learning Environments, 29(1), 142–163.
Nguyen, N. D. (2023). Exploring the role of AI in education. London Journal of Social Sciences, (6), 84–95. https://doi.org/10.31039/ljss.2023.6.108
Organización de las Naciones Unidas para la Educación, la Ciencia y la Cultura (Unesco). (2019). Artificial intelligence in education: Challenges and opportunities for sustainable development. https://unesdoc.unesco.org/ark:/48223/pf0000366994.locale=es
Padilla Escorcia, I. A., & Conde-Carmona, R. J. (2020). Uso y formación en TIC en profesores de matemáticas: un análisis cualitativo. Revista Virtual Universidad Católica del Norte, (60), 116–136. https://revistavirtual.ucn.edu.co/index.php/RevistaUCN/article/view/1166
Panqueban, D., & Huincahue, J. (2024). Inteligencia artificial en educación matemática: una revisión sistemática. Uniciencia, 38(1), 1–17. http://dx.doi.org/10.15359/ru.38-1.20
Patton, M. Q. (2015). Qualitative research & evaluation methods: Integrating theory and practice (4th ed.). Sage Publications.
Rasheed, N. M., & Tashtoush, M. A. (2023). The impact of Cognitive Training Program for Children (CTPC) to development the mathematical conceptual and achievement. Journal of Higher Education Theory and Practice, 23(10), 218–234. https://doi.org/10.33423/jhetp.v23i10.6196
Shin, W.-S., & Shin, D.-H. (2020). A study on the application of artificial intelligence in elementary science education. Journal of Korean Elementary Science Education, 39(1), 117–132. https://doi.org/10.15267/keses.2020.39.1.117
Shirawia, N., AlAli, R., Wardat, Y., Tashtoush, M. A., Saleh, S., & Helali, M. (2023). Logical mathematical intelligence and its impact on the academic achievement for pre-service math teachers. Journal of Educational and Social Research, 13(6), 242–257. https://doi.org/10.36941/jesr-2023-0161
Tashtoush, M. (2019). Weakly c–normal and cs–normal subgroups of finite groups. Jordan Journal of Mathematics and Statistics, 1(2), 123–132.
Van Manen, M. (2016). Phenomenology of practice: Meaning-giving methods in phenomenological research and writing. Routledge. https://doi.org/10.4324/9781315422657
Wang, S., Yu, H., Hu, X., & Li, J. (2020). Participant or spector? Comprehending the willingness of faculty to use intelligent tutoring systems in the artificial intelligence era. British Journal of Educational Technology, 51(5), 1657–1673. https://doi.org/10.1111/bjet.12998
........................................................................................................................................................
Robinson Junior Conde-Carmona
PhD in Mathematics Education from Antonio Nariño University. Professor at the Faculty of Education and a researcher in the GIMED group at Universidad del Atlántico, Colombia. He is a research associate classified by the Ministry of Sciences. His research centers on mathematics education, teacher training, and the integration of technologies in teaching. His recent publications include the articles “Caracterización del conocimiento especializado del profesor de matemáticas en la enseñanza de las fracciones,” co-authored in Encuentros, 22(01), 98–113, http://ojs.uac.edu.co/index.php/encuentros/article/view/3097; and “STEAM para el desarrollo del pensamiento matemático: una revisión documental,” co-authored in Praxis, 20(2), https://doi.org/10.21676/23897856.5783.
Iván Andrés Padilla-Escorcia
PhD (c) in Sciences in the field of Educational Mathematics. Professor at the Faculty of Education and researcher at the GIMED group at Universidad del Atlántico, Colombia. His areas of interest include specialized knowledge of mathematics teachers, teaching mathematics through technology, and mathematical modeling. His recent publications include articles. “Caracterización del conocimiento especializado del profesor de matemáticas en la enseñanza de las fracciones,” co-authored in Encuentros, 22(01), 98–113, http://ojs.uac.edu.co/index.php/encuentros/article/view/3097; and “Specialised Knowledge of the Mathematics Teacher to Teach through Modelling using ICTs” (2023), co-authored in Acta Scientiae, 25(1), 160–195, https://doi.org/10.17648/acta.scientiae.7363.
* This article is part of a project on mathematics education related to AI. It was not funded, and there are no conflicts of interest to disclose. The authors’ contributions to the writing are as follows: Robinson Junior Conde-Carmona was responsible for the introduction, construction of the theoretical framework, interpretation, analysis, and writing; and Iván Andrés Padilla-Escorcia for the methodology, editorial review, and organization of the bibliography. Correspondence regarding this work should be sent to Robinson Junior Conde-Carmona ( rjconde@mail.uniatlantico.edu.co ). The article was translated with funding from the Vice President for Research and Creation at Universidad de los Andes (Colombia). This article was first published in Spanish as: Conde-Carmona, R. J., & Padilla-Escorcia, I. A. (2025). La inteligencia artificial en educación matemática: percepciones de futuros docentes sobre expectativas, prácticas y desafíos. Voces Y Silencios. Revista Latinoamericana De Educación, 16(2), 1-26.