The chosen databases were Scopus and Science Direct based on their reputation for publishing quality research papers in the area of technology for math education. Scopus ensured access to a broad spectrum of research articles relevant to the intersection of technology and math education, while science direct would potential provide full-text access to a large collection of scientific and technical research articles, particularly strong in fields like educational technology and STEM education, making it highly relevant for the study. A total of 423 papers were retrieved from these databases on the 16th December 2023. The following search phrases were used to retrieve the research papers: (“Innovative Technology” OR “Enabling Technology”) AND (“Math Instruction” OR “Mathematics Education” OR “Math teaching and learning”).

The screening was based on the inclusion/exclusion criteria shown in Table 1 . The initial screening identified 77 articles which were not accessible, 16 duplicates and 9 non-English papers. Thereafter, the remaining 321 sources were screened to ensure that they met our inclusion criteria. This process resulted in the exclusion 68 editorials, opinion pieces or content that did not directly address the research question.

Thereafter, the remaining 253 papers were assessed for quality using the following quality assessment (QA) criteria:

1) Does the paper discuss the different dimensions or components of the technologies adopted for math education?

2) Does the paper discuss how technologies could impact the potential transformation in math education?

Each Quality Assessment question carried a score of 1. Based on criterion 1, papers that clearly detailed the technological aspects received a score of 1, while those that lacked this information scored 0. In line with criterion 2, papers that provided a thorough analysis of the potential transformative effects of technology in math education received a score of 1, whereas papers that did not address this aspect scored 0. Thus, a paper could have a maximum score of 2. Only sources that obtained at a score of at least 1 were considered for inclusion in the SLR.

Following the QA process, a total of 117 papers were retained for the SLR. These papers met the minimum quality threshold and were considered to provide valuable contributions to the understanding of how innovative technologies can enhance math education. 63 papers obtained a score of 1 while 54 obtained a score of 2. Figure 1 shows the number of papers selected for the automated content analysis by year. It is observed that the majority of them have been published for the last 5 years suggesting a relatively recent emergence of these technologies.

The review was conducted using the Leximancer which is a software package developed for performing conceptual analysis of natural language textual data and was conducted on the basis of the four stages of the Leximancer project control, namely “select documents, generate concept seeds, generate thesaurus and generate results.” For the “select document” stage, the 117 papers were included. The second step was to initiate the generation of concept seeds. During this step, the software also executes the stemming task (reducing words to their base form) and removes common stop words (e.g., “and,” “the,” “is” “paper”, “study”) to focus on more meaningful terms. In the third step, Leximancer identifies concepts and relationships within the text, based on different algorithms. In the last step, Leximancer provides interactive visualization tools such as concept maps to analyze the results ( Leximancer, 2021 ).

Out of the 117 selected papers, Leximancer generated a total of 7 themes with their corresponding concepts. The themes with the most significant number of hits are “student”, “Learning”, “mathematics”, “model”, “Intelligence”, “ChatGPT” and “Technology” as shown in Figure 2 which also showcases the concept map for the 7 themes. Furthermore, Table 2 lists the main concepts associated with the corresponding themes.

Source: Authors’ own compilation.

Technology integration models

Wachira & Keengwe (2011) define technology integration as “incorporating technology and technology-based practices into all aspects of teaching and learning and assessments.” The literature suggests four important technology integration models namely, the Technological Pedagogical Content Knowledge (TPCK), the Technological Acceptance Model (TAM), the Unified Theory of Acceptance and Use of Technology (UTAUT) and the Technology-Organization-Environment (TOE). These models delve into several aspects of technology integration and adoption including acceptance, knowledge, will, belief and performance. Guerrero (2010) showed that TPCK embodies a teacher’s ability to distinguish between the types of technology that are most suited to their mathematical content and instructional goals and make decisions regarding its appropriate application. The Technology Acceptance Model (TAM) is the key model uncovering behavioural patterns for accepting or rejecting technology in the area of teaching and learning. In the context of mathematics education, learning conditions, parental involvement, school curriculum and teacher training, emerge as factors contributing to acceptance ( Arthur, 2022 ). The UTAUT comprises 4 main components namely, social influence, performance expectancy, effort expectancy and enabling conditions ( Venkatesh et al., 2003 ). The TOE on the other hand explains the influence of technology, organizational attributes, and environmental factors in the process of implementation and adoption. ( Baker, 2012 )

Technology for Math Instruction

Learning mathematical procedures is important, but it is also crucial to equip students with strong problem-solving and reasoning, through mathematical skills acquisition ( Lake et al., 2017 ). Kynigos (2019) and Sriraman & English (2010) stipulate that within each educational institution, a model needs to be built and sustained on how to use technology to make math instruction more effective. Sacristán (2017) consider the diverse elements of the development model: training, continuous support, processes of reflection, self-observation and point out to the fact that promoting equally the technological, mathematical and pedagogical aspects, help to construct didactic strategies in accordance with the local educational practices. Studies of Hoyles et al. (2020) and Trinh Thi Phuong et al. (2022) lay emphasis on the integration of a wide range of culturally relevant digital tools for teaching mathematics, as well as the importance of teacher training for an effective and strategic use of those resources in the classroom.

Leveraging Technology for Teacher support

Technology contributes to developing teachers’ responsive teaching skills and instructional methods to support the development of the learners’ skills in presenting their mathematical thinking ( Hoyles et al., 2020 ; Kaitera & Harmoinen, 2022 ). Advances in Artificial Intelligence (AI) and Natural Language Processing (NLP) support teachers with automated feedback on their use of classroom discourse. Applications building on advances in deep learning for natural language processing and speech recognition analyze classroom recordings and provide math teachers with personalized feedback on their instructional practices ( Suresh, 2022 ). According to Sinclair and Yerushalmy (2016) , classroom response systems are participatory and include feedback tools to enhance the students’ engagement in math learning. Technology helps teachers support groups of students by stimulating mathematical discussions and enhance classroom communication ( Calor et al., 2022 ). Furthermore, Schwarz et al. (2021) stress on the adaptive guidance through dashboards to provide effective support for teachers to interact with multiple small groups of students, collaborating on mathematics. Nevertheless, it is imperative to acknowledge the limitations of these technological tools for math teaching and learning and use them in a way that complements the human educators rather than replacing them ( Atlas, 2023 ; Viberg, Gronlund, & Andersson, 2023 ). Sacristán (2017) uncovers the difficulties and obstacles teachers and students face while integrating the technological tasks with the established curriculum. These are often due to limited time and resources for the pedagogical content preparation, and implementation. Trigueros & Sacristan (2008) establish that the use of technological tools for understanding of mathematical concepts, does not necessarily lead to increase achievement, particularly because much of the knowledge developed can remain situated in the technological context.

Student Centered Learning

Students are at the center of all teaching and learning activities. A broad-based approach is fundamental to ensure that the math learning activities optimally engage the student through adequate knowledge representations coupled with experiential, contextual, and collaborative learning ( Lai & Cheong, 2022 ). Woolf (2010) states that intelligent tutors demonstrate their effectiveness in a broad range of applications which can infer student learning strategies. For example, machine learning enables tutors to reason about uncertainty and predict performance, based on observed student behavior. In addition, chatbots can also be used to track student progress and provide data to teachers and administrators about student performance. This approach is aligned with the larger framework of external accountability, which places a greater emphasis on measuring student achievement and holding schools and teachers accountable for their performance ( Birenbaum, 2023 ). Calor et al. (2022) analyse classroom response systems which are participatory and include feedback tools that can increase the students’ engagement in math learning and enhance classroom communication between teacher and students. Kim et al. (2020) highlight significant implications for instructional communication for math related content. While technological skills are well advanced to create machine teachers, there is little understanding about how students would perceive the idea of machine teachers especially for technical mathematical content ( Azevedo et al., 2022 ). The focus is nonetheless on developing an understanding of “students’ mathematics and students’ mathematical realities.” ( Schoenfeld, 2010 )

Personalised Learning with Technology

According to Ahmad et al. (2023) the traditional classroom teaching and learning in Education 1.0 was a one-size-fits-all approach which did not consider the individual needs of students. This made it difficult for teachers to provide personalized instruction and support to students with different learning styles and abilities. Personalised teaching and learning using technological tools such as intelligent tutoring systems help them adapt their content support to individual students’ different levels of understanding which can be heterogeneous in math classrooms. Hwang & Tu (2021) review the adaptive systems to provide personalized guidance to almost all levels of students by analysing their learning status and behaviors. They point out to the fact that in addition to mathematical skills, motivational aspects play important roles in an individual’s capacity in mathematical problem solving situations. Communication and collaboration with peers are also essential for expressing mathematical ideas ( Kaitera & Harmoinen, 2022 ). Additionally, the use of technology such as game-based learning can promote problem-solving and critical thinking skills, which are important 21st-century skills. Furthermore, game-based learning can be personalized to the needs of individual students, allowing them to work at their own pace, which can help improve learning outcomes for all students ( Ahmad et al., 2023 ). This in turn contributes to build up students’ mathematical self-confidence in describing their problem-solving processes. As a result, students gain a deeper understanding of mathematical ideas ( Hwang, Flavin, & Lee., 2023 ). Nevertheless, simply using technology is not enough: technology must be properly integrated with work techniques, curriculum, learning and mathematical assessment ( Cahyono & Asikin, 2019 ).

Artificial Intelligence (AI) in Math Education

According to Chen, Chen, & Lin (2020) , AI induces instructional quality and effectiveness in the contemporary technology-based adaptive teaching and learning systems. These foster customization and personalized content in line with the learner capabilities and needs, thereby improving learners experience and contributing to the achievement of the set learning objectives. Automation and AI-based tools have the potential to support teachers in improving their pedagogical skills and increasing their self-efficacy ( Suresh 2022 ). Studies such as Lee & Perret (2022) and Kasneci et al. (2023) indicate a positive attitude towards AI for education and a high motivation to introduce AI-related content at school. Teacherbots emerge as disruptive alternative to traditional teaching staff ( Popenici & Kerr, 2017 ). AI-generated strategies provide teachers with step-by-step explanations of math procedures to guide their thinking as they plan and implement individual learning experiences for students ( Gattupalli, Maloy, & Edwards, 2023 ). Leveraging and using AI in education has fostered effectiveness and efficiency in the performance of administrative tasks, such as grading of students’ assignments. AI has significantly reduced the paperwork and workload on instructors, enabling them to focus on their core mandate ( Russell & Norvig, 2010 ). Chen, Chen & Lin (2020) highlight the different pedagogical tools that incorporate and leverage AI into web-based education. Teachers utilize chatbots to design tutoring interventions and assessment tools customized to students’ different preferences and levels, give feedback on students’ performances, and stimulate students’ curiosity and higher-order thinking. As reported by Luckin and Holmes (2016) , intelligent support for teachers could also help address the issue of teacher retention in a field where many skilled professionals are leaving due to “burnout”.

ChatGPT and other language models can be useful tools for teaching and learning Mathematics. They can assist human educators in their work by providing additional resources and support ( Atlas, 2023 ). ChatGPT is a chatbot recognized for its improved math capabilities and ability to offer comprehensive instruction and increase educational understanding of mathematical concepts. ( Wardat et al., 2023 ). This tool can be used for incorporating critical thinking and problem-solving activities into the curriculum, and help students develop these skills. However, educators should provide training and support to students on how to use ChatGPT responsibly, including proper attribution and ethical considerations. It is crucial to respect the academic integrity policies of the institution ( Atlas, 2023 ). ChatGPT is nevertheless not capable of replacing the unique skills and abilities of human educators to review, validate and explain the mathematical concepts. It is important to use these tools in conjunction with human instruction and support and be aware of their limitations for math education ( Baidoo-Anu & Ansah, 2023 ). Birenbaum (2023) refers to ChatGPT’s lack of emotional intelligence whereas human teachers bring creativity, critical thinking skills, and emotional intelligence to the classroom that cannot be replicated by chatbots.