Empowering Mathematics Educators: Integrating ChatGPT as a Tool

for Innovative Teaching Practices

Maria Lucia Bernardi

, Roberto Capone

and Mario Lepore

Department of Mathematics, University of Bari Aldo Moro, Orabona Street, Bari, Italy

Department of Mathematics, University of Salerno, Fisciano (SA), Italy

Keywords: Artificial Intelligence, Teachers’ Education, Function Continuity, ChatGPT.

Abstract: This study investigates the potential of a customized ChatGPT model as a tool for enhancing mathematics

teaching, specifically focusing on the concept of continuity for real-valued functions. Using the Knowledge

for Teaching Mathematics with Technology (KTMT) framework as a theoretical basis, the research examines

how personalized AI tools can improve task design, balance among mathematical representations, and the

interplay between experimentation and justification. The experimentation involved in-service mathematics

teachers who explored both a default and a customized ChatGPT model to create instructional resources.

Qualitative analysis revealed that the customized model significantly improved the quality of resources,

enabling the creation of diverse, representation-rich, and conceptually balanced tasks. Teachers reported that

the personalized ChatGPT facilitated transitions between algebraic, graphical, and tabular representations,

supported exploratory problem-solving, and provided opportunities for rigorous justification. The findings

contribute to a deeper understanding of how AI-driven tools, when aligned with structured pedagogical

frameworks, can support mathematics instruction and teacher development.

1 INTRODUCTION AND

RATIONALE

The integration of digital technologies in

mathematics education presents both challenges and

opportunities for teachers' professional development

(Drijvers and Sinclair, 2023). Artificial Intelligence

(AI), particularly tools like ChatGPT, offers the

potential to enhance teaching by enabling

customizable and interactive learning resources,

though concerns about reliability and accuracy

remain (Bernardi et al., 2024a). The effectiveness of

these tools depends on teachers’ ability to use them

strategically, which requires targeted training.

This study applies the Knowledge for Teaching

Mathematics with Technology (KTMT) framework

(Rocha, 2020) to analyze how AI can be leveraged to

support mathematics instruction, specifically

focusing on the concept of continuity for real-valued

functions. The KTMT framework identifies three key

dimensions:

https://orcid.org/0009-0001-1851-7361

https://orcid.org/0000-0001-9858-9420

https://orcid.org/0000-0001-9454-8453

Task Characteristics: Attention to the cognitive

demand, the level of structuring, and the role of

technology in the proposed tasks.

Balance Among Mathematical Representations:

Effective articulation and integration of diverse

representations, such as algebraic, graphical, and

numerical formats.

Experimentation and Justification: A balance

between leveraging technology for experimentation

and fostering mathematical justification through

argumentation and proof.

In this research, these dimensions serve as a basis

to investigate the impact of using ChatGPT in

mathematics education. Specifically, the study

focuses on teaching the concept of continuity for real-

valued functions of a real variable, a central topic for

fostering students’ analytical thinking.

In this study group of in-service teachers initially

used ChatGPT in its default state to generate

instructional materials and then experimented with a

customized version designed to produce

404

Bernardi, M. L., Capone, R. and Lepore, M.

Empowering Mathematics Educators: Integrating ChatGPT as a Tool for Innovative Teaching Practices.

DOI: 10.5220/0013437900003932

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Conference on Computer Supported Education (CSEDU 2025) - Volume 2, pages 404-411

ISBN: 978-989-758-746-7; ISSN: 2184-5026

pedagogically refined content (Bernardi et al.,

2024b).

The research seeks to validate whether the

potential identified by researchers aligns with

teachers’ experiences. To this end, the study is guided

by two primary research questions:

Does using a customized ChatGPT model

personalized to the topic of continuity provide

benefits to teachers compared to using the default,

non-customized version?

How do the three key dimensions identified by the

KTMT framework—task characteristics, balance

among representations, and the interplay between

experimentation and justification—manifest and

evolve when utilizing the customized ChatGPT?

The analysis is based on qualitative data collected

through an interpretive observational study,

employing tools such as in-depth interviews to

explore the perceptions, experiences, and strategies of

participating teachers.

The scientific contribution of this research lies in

demonstrating how the KTMT framework can be

employed to analyze the potential of AI in

mathematics education, providing a structured

approach to assessing its pedagogical impact. Unlike

previous works that examine ChatGPT’s role through

general pedagogical models, such as the didactical

tetrahedron (Dasari et al., 2024), dialogic learning

(Pavlova, 2024), and the Zone of Proximal

Development (Govender, 2023), this study

specifically situates AI’s impact within a framework

designed to assess teachers’ technological knowledge

in mathematics instruction. The findings aim to offer

practical insights for teacher training programs,

supporting a more effective and reflective integration

of AI into mathematics education.

2 THEORETICAL FRAMEWORK

As a theoretical framework, we use The Knowledge

for Teaching Mathematics with Technology (KTMT).

It is a framework conceptualizing teachers’

professional knowledge to integrate research on

teacher knowledge and the incorporation of digital

technology (DT) into teaching practice (Rocha, 2013).

It builds on foundational work by Shulman (1986) and

Mishra and Koehler (2006) and identifies three

categories of knowledge: base knowledge domains,

inter-domain knowledge, and integrated knowledge.

The base knowledge domains include Mathematics,

Teaching and Learning, Technology, and Curriculum

and Context. These domains align with those found in

established teacher knowledge models, but KTMT

uniquely positions the Curriculum and Context

domain as transversal, influencing all other domains.

A distinguishing feature of KTMT is its two inter-

domain knowledge sets:

Mathematics and Technology Knowledge (MTK):

Focuses on how DT shapes mathematical concepts,

either enhancing or constraining particular aspects.

Teaching and Learning and Technology

Knowledge (TLTK): Examines how DT influences

teaching and learning strategies, similarly, acting as

either an enabler or a limitation.

Both MTK and TLTK draw on research into the

integration of DT, identifying its impacts on task

design, teaching approaches, and mathematical

exploration. For example, DT facilitates exploratory

tasks where students analyze cases, formulate

conjectures, and identify patterns, thereby enriching

the teaching and learning process (Goos and

Bennison, 2008). DT also offers access to diverse

representations, requiring teachers to guide students in

interpreting and articulating the information across

these formats (Rocha, 2013). Finally, KTMT

introduces Integrated Knowledge (IK)—a

comprehensive synthesis of the base domains and

inter-domain knowledge. IK reflects the teachers’

ability to interweave all these elements to effectively

integrate DT into their practice, ensuring it supports

both mathematical understanding and pedagogical

goals.

Figure 1: KTMT model and the three dimensions.

In Fig. 1, the painter’s metaphor is used to

represent the KTMT model. The pallet represents the

knowledge of the curriculum and of the context in

which the teacher works. The three primary colors

(red, blue, and yellow) represent the base knowledge

domains (Mathematics, Teaching and Learning, and

Technology). The three secondary colors (orange,

green, and purple) represent MTK, TLTK, and PCK

(Pedagogical Content Knowledge). Each of these is a

new color (new knowledge) created by the merging of

Empowering Mathematics Educators: Integrating ChatGPT as a Tool for Innovative Teaching Practices

405

two primary colors and assuming different tonalities

depending on the amount of each of the primary colors

mixed (different teacher knowledge). This metaphor

highlights differences from the idea of intersection

present in TPACK, where the development depends

directly on the knowledge in the base domains (Mishra

and Koehler 2006). In KTMT, an increase in the

amount of ink of one primary color may or may not

result in a tonality change of the secondary color. The

result depends on the quantity of the primary color that

is actually mixed with some other primary color (an

increase in the knowledge of one of the base

knowledge areas need not result directly in an increase

of related inter-domain knowledge).

This study adopts a theoretical framework based

on three dimensions critical to teaching mathematics

with digital technology (DT) in our study ChatGPT,

derived from the Knowledge for Teaching

Mathematics with Technology (KTMT) model:

Type of Task and Use of Technology: DT enables

diverse tasks (e.g., problems, investigations,

modeling) that vary in structure and cognitive

demand. Teachers must design tasks that leverage DT

affordances, balancing procedural and conceptual

approaches while focusing on the intended learning

outcomes (Ponte, 2005).

Representations and Their Use: DT facilitates

dynamic linking of representations (algebraic,

graphical, numeric, and tabular), enhancing

conceptual understanding (Duval, 2006). Teachers

must balance and articulate transitions between

representations to maximize learning and avoid

privileging specific forms over others (Rocha, 2013).

Experimentation and Justification: DT fosters

experimentation but can lead to over-reliance on

empirical evidence. Teachers must guide students to

move beyond conjectures to deductive reasoning,

balancing exploration with rigorous justification to

highlight the role of proof in mathematics (Lesseig,

2016).

These dimensions provide a lens to examine how

teachers integrate DT effectively in mathematics

instruction, focusing on task design, representation

use, and the interplay between exploration and

justification.

3 METHODS

3.1 Experimental Design

The experiment was conducted with a group of 15 in-

service mathematics teachers who had limited prior

experience with ChatGPT but demonstrated a strong

interest in exploring innovative teaching practices for

their classrooms. The study was designed as an

exploratory investigation, organized into distinct

phases to evaluate the potential of ChatGPT as a

teaching tool. The primary objective was to compare

the efficacy of the default ChatGPT with a version

specifically personalized for teaching the concept of

continuity in real-valued functions.

The experiment comprised the following phases:

Phase 1 (P1): exploration of ChatGPT without

instructions;

Phase 2 (P2): exploration of ChatGPT with

instructions;

Phase 3 (P3): reflections and final interview.

In the first phase, participants interacted with

ChatGPT in its default configuration, without any

prior customization or specific instructions. The focus

was on the topic of continuity for real-valued

functions of a real variable. Teachers explored the

system independently, asking questions and

generating resources as needed. The primary aim of

this phase was to observe how the teachers engaged

with the tool in its generic state and to analyze the

types and quality of the outputs generated in response

to their queries.

In the second phase, participants were introduced

to a customized version of ChatGPT, specifically

instructed on the topic of continuity for real-valued

functions. This personalized version was accessible

via a dedicated link and was designed to provide

explanations, examples, and tasks aligned with

pedagogical goals. The purpose of this phase was to

assess how the customization influenced the

relevance, accuracy, and educational value of the

outputs compared to the first phase. Teachers were

encouraged to explore the model’s capabilities for

generating personalized resources and to reflect on its

potential as a tool for supporting mathematical

reasoning and instruction.

The final phase involved a structured reflection

and feedback session, providing insights into the

perceived strengths and limitations of both versions

of ChatGPT. They discussed possible applications of

the tool in their teaching practices and proposed

recommendations for its integration into mathematics

education.

3.2 Setup and Configuration of the

Continuity-Specific ChatGPT

As part of this research, a specialized ChatGPT model

was developed to enhance the exploration and

teaching of the concept of continuity for real-valued

functions. This tool was designed not only to support

CSEDU 2025 - 17th International Conference on Computer Supported Education

406

learners but also to inspire educators in designing

innovative teaching strategies. The creation of this

customized version of ChatGPT required a carefully

planned process to ensure that it could address the

specific needs of mathematics education in this

domain.

The first step in configuring the model was to

define its objectives clearly. The idea was to create a

conversational assistant capable of engaging users

with clear and adaptable explanations, solving

mathematical problems related to continuity, and

providing pedagogical insights for teaching this

fundamental concept. To achieve this, a dataset was

prepared containing theoretical content, solved

examples, practical exercises, and real-world

applications. This dataset covered topics such as the

definition of continuity, types of discontinuities (e.g.,

removable, jump, and infinite), the relationship

between continuity and limits, and graphical

interpretations of these concepts. To make the tool

versatile, examples were included with varying levels

of difficulty, and the responses were designed to be

dynamic, encouraging users to explore deeper

questions and scenarios.

The model was fine-tuned using this dataset to

ensure that its responses were accurate, contextually

appropriate, and reflective of the pedagogical goals of

the project. Special attention was given to the way the

model interacts with users: it was designed to adapt

to their level of knowledge, providing simpler

explanations for beginners and diving into more

complex details for advanced users. For instance,

when asked about the continuity of a function, the

model might start with a basic definition and then

guide the user through a step-by-step analysis of a

practical example, incorporating graphical

visualizations when relevant.

To make the interaction engaging and inspiring,

the model includes elements that stimulate curiosity

and real-world connections. For example, it might ask

users reflective questions like, “Have you ever

wondered how the concept of continuity is used to

ensure smooth animations in digital graphics?” or

propose links to everyday applications of continuity

in physics or engineering. Additionally, it was

designed to provide motivational feedback, such as

highlighting the user’s strengths in problem-solving

while suggesting areas for improvement. The tone of

the interaction was carefully calibrated to be

conversational, encouraging, and inspiring,

leveraging emojis and thoughtful phrasing to

maintain engagement.

The specialized ChatGPT is equipped to perform

a range of functions adapted to the needs of both

learners and educators. These include responding to

theoretical questions about continuity, solving

problems step by step, generating exercises with

progressive difficulty, and providing personalized

feedback on user-generated solutions. For educators,

the tool offers suggestions for teaching strategies and

classroom activities, such as designing exercises that

link continuity to real-world phenomena or using

graphical representations to explain abstract

concepts.

This tool is available at the following link:

https://chatgpt.com/g/g6759b2b8242c8191b8243f3f

ba7099d6-continuita-funzioni-reali-gpt.

Educators and students alike are encouraged to

explore the potential of this specialized ChatGPT.

Whether it’s generating interactive lesson plans,

testing one’s understanding with custom exercises, or

delving into the philosophical beauty of mathematical

continuity, this tool is designed to be a versatile and

inspiring companion. By combining adaptability,

interactivity, and pedagogical value, it represents a

step forward in integrating artificial intelligence into

mathematics education, creating new possibilities for

learning and teaching alike.

3.3 Data Collection Instruments

To ensure a comprehensive analysis of the teachers’

experiences and perceptions about ChatGPT’s

usefulness and potential in mathematics education

qualitative observations and in-depth interviews were

employed. The interactions of teachers with ChatGPT

during the sessions were observed, focusing on their

problem-solving strategies, patterns of tool usage,

and engagement with the AI.

3.4 Data Analysis

The collected data were analyzed using a qualitative

approach, focusing on identifying themes and

patterns related to the use of ChatGPT in mathematics

education, taking into account the three dimensions

critical to teaching mathematics with digital

technology (DT), derived from the Knowledge for

Teaching Mathematics with Technology (KTMT)

model.

4 RESULTS AND DISCUSSION

The findings of our study highlight that the use of the

customized ChatGPT had a significant impact on the

quality of the instructional materials produced by

teachers. This customization contributed to the

Empowering Mathematics Educators: Integrating ChatGPT as a Tool for Innovative Teaching Practices

407

development of more personalized and innovative

lessons, particularly in teaching the concept of

continuity for real-valued functions. However,

participants’ comments during the initial phase

reflected concerns about the reliability and

appropriate use of such a tool. For instance, T1 noted,

“I am very doubtful about the correctness of

ChatGPT’s responses; I fear it might generate errors

or inaccurate content.” Similarly, T2 remarked, “I

think that the use of AI will, over time, further

diminish the teacher’s central role in mediating the

learning process.” T3 expressed uncertainty about

their ability to fully utilize ChatGPT, stating, “I don’t

know if I am capable of using ChatGPT to its full

potential; without proper training, I might cause more

harm than good, and I don’t think it is particularly

intuitive to use.”

These concerns were partially mitigated as the

experiment progressed and participants explored the

potential of the customized tool. Below are excerpts

from an in-depth interview with three teachers who

voluntarily consented to continue the discussion

beyond the main phases of the study.

Do you think ChatGPT can help you design

innovative lessons on the continuity of functions?

T1: ChatGPT allows me to design innovative

lessons on continuity that balance experimentation

and justification, stimulate transitions between

different representations, and offer customizable

tasks with strong educational value.

In what ways?

T1: ChatGPT enables me to create diverse tasks

with varying levels of structure and cognitive

complexity. I can use it to design exploratory

problems where students verify the continuity of a

function through real-world scenarios or practical

applications. For instance, I discovered that ChatGPT

can generate examples of functions representing

natural or economic phenomena, allowing students to

model and analyze data. This approach supports a

balance between procedural tasks, like calculating

limits, and conceptual understanding, such as

distinguishing between pointwise and global

continuity. Additionally, the chatbot helps me

customize activities based on students’ specific

needs, generating tasks that encourage different levels

of mathematical reasoning.

T2: ChatGPT is particularly useful for creating

activities that integrate multiple representations of

functions. I asked the chatbot to provide a textual

description of a function accompanied by graphical

representations, value tables, and algebraic

formulations. This capability allows me to encourage

dynamic transitions between different forms of

representation, which is critical for fostering a deep

understanding of continuity. Moreover, the chatbot

suggested innovative ways to link these

representations, ensuring that students don’t focus

exclusively on one form—such as graphs—while

neglecting symbolic reasoning.

T3: ChatGPT facilitates the exploration of

hypotheses about continuity by enabling dynamic

manipulation of parameters in a function and

connecting with tools like GeoGebra for interactive

simulations. When teaching discontinuities, I can use

ChatGPT to generate examples of functions with

removable, jump, or essential discontinuities,

describing their algebraic characteristics. For

example, the chatbot can define and illustrate a

function with a jump discontinuity, such as 𝑓



𝑥











, generate the associated graph, and visually

highlight where the discontinuity occurs. This helps

students connect the symbolic form of the equation

with the visual behavior of the function.

Have you noticed differences between responses

from the instructed chatbot versus the default

version?

T3: Yes, especially when discussing the three

types of discontinuities. I documented these

differences with screenshots.

T2: One critical advantage of the instructed

ChatGPT is its ability to create immediate

connections between different representations. For

instance, after presenting a function algebraically, I

can ask the chatbot to provide textual descriptions

explaining the meaning of the discontinuity in

language accessible to students. It can then integrate

graphs and tables to demonstrate how the

discontinuity emerges across various representational

contexts, reinforcing the mathematical phenomenon.

T1: I made this request to ChatGPT: “After

presenting a function algebraically, could you create

textual descriptions that explain the meaning of the

discontinuity using language accessible to students?”

This was the response from the “non-instructed”

chatbot regarding jump discontinuities:

CSEDU 2025 - 17th International Conference on Computer Supported Education

408

Table 1: “Non-instructed” ChatGPT response and

translation in English.

1. Jump Discontinuity

Imagine walking along the graph of a function. A

jump discontinuity is like an unexpected step: as

you walk along the curve, you arrive at a point

where you must "jump" to continue. This happens

when two pieces of the graph, one to the left and

one to the right of the point, do not connect. For

example, if you observe the point x=1 in the

function f(x), you might notice that the curve

"jumps" from one value to another without passing

through the middle. This means that the left-hand

limit (lim

→

f(x) ) is different from the right-hand

limit (lim

→

f(x) ).

T1: This is the response from the instructed

chatbot:

Table 2: “Instructed” ChatGPT response and translation in

English.

User: Afte

resenting an algebraic function, could

you create textual descriptions that explain the

meaning of discontinuity, using language

accessible to students?

ChatGPT: Absolutely! I will provide an example

for each type of discontinuity using an algebraic

function, followed by a textual description

accessible to students. Let’s

started!

1. First Kind Discontinuit

(Jump

Discontinuity)

Function:

Explanation:

This function has a first kind discontinuity at x=1,

because:

1. lim

→

f(x)=2(1) +1 =3

2. lim

→

f(x)=3(1) - 4 =-1.

The left-hand limit and right-hand limit exist but

are different from each other. This creates a

"jump" in the graph at x = 1. Imagine drawing the

graph

—

there's an interruption, like a "step."

T1: It also provided a metaphor for the students:

“ Metaphor for Students: Imagine you’re walking

along a straight road, but suddenly you find a step

change: you have to jump to keep going. This jump

represents the discontinuity! 󰼳󰼴󰼵󰼶󰼷󰼸󰼹󰼺󰼻”

Does the use of ChatGPT diminish the teacher’s

role in the classroom?

T1: My role remains crucial for guiding students

toward a critical use of technology and consolidating

the mathematical rigor required for learning.

T3: One of the main risks, in my view, is over-

reliance on empirical verification. For instance,

students might conclude that a function is continuous

based solely on observing its graph without formally

proving it. To avoid this, I would integrate ChatGPT-

generated resources into activities that require

students to formulate conjectures through exploration

but subsequently justify them with rigorous

deductions using definitions and theorems. This way,

I can balance the exploratory aspect provided by

technology with the rigor of mathematical reasoning.

The results of this study demonstrate that the

instructed ChatGPT significantly improved the

quality of instructional resources produced by

teachers, addressing the three key dimensions

identified by the KTMT framework: task

characteristics, balance among mathematical

representations, and the interplay between

experimentation and justification. These findings

underscore the potential of personalized AI tools to

align with and enhance pedagogical goals in

mathematics education.

From the perspective of task characteristics, the

instructed ChatGPT enabled teachers to design tasks

that were not only diverse but also aaptable to the

Empowering Mathematics Educators: Integrating ChatGPT as a Tool for Innovative Teaching Practices

409

cognitive levels of their students. This aligns with the

KTMT framework's emphasis on leveraging digital

tools to create tasks with varying degrees of structure

and cognitive demand. For instance, teachers used the

instructed ChatGPT to generate exploratory problems

that required students to verify the continuity of a

function in real-world scenarios, such as modeling

natural phenomena or economic trends. These tasks

encouraged critical thinking by connecting abstract

mathematical concepts to practical applications,

fostering a deeper understanding of continuity

beyond procedural fluency. The personalized model

also allowed for incremental scaffolding, enabling

teachers to adjust task complexity dynamically. This

reflects the importance of technology in supporting

differentiated instruction, as highlighted by the

KTMT model.

In terms of balance among mathematical

representations, the instructed ChatGPT

demonstrated its ability to seamlessly integrate

algebraic, graphical, and tabular formats. This

capability resonates with the KTMT framework’s call

for dynamic linking of representations to enhance

conceptual understanding. Teachers reported that the

tool facilitated transitions between these formats,

helping students explore the interconnectedness of

different representations of continuity. For example,

one teacher highlighted how the chatbot could

generate a textual explanation of a function’s

discontinuity alongside its graph and tabular

representation, ensuring that students did not overly

rely on one form, such as visual graphs, while

neglecting symbolic reasoning. This balance is

critical for promoting a holistic comprehension of

mathematical concepts, a challenge often observed in

traditional teaching practices.

The interplay between experimentation and

justification was another area where the instructed

ChatGPT made a meaningful impact. While digital

tools like ChatGPT naturally encourage exploratory

learning through simulations and dynamic parameter

adjustments, the KTMT framework stresses the need

for rigorous justification to prevent over-reliance on

empirical observations. Teachers noted that the

instructed ChatGPT provided tasks that encouraged

students to formulate hypotheses about continuity but

also required them to validate their conjectures using

formal definitions and proofs. For example, teachers

used the tool to guide students through analyzing

removable and jump discontinuities, transitioning

from initial observations to deductive reasoning. This

dual focus on experimentation and justification not

only deepened students’ understanding of continuity

but also reinforced the importance of mathematical

rigor.

Reflecting on the two research questions posed at

the outset, the study provides clear answers. First, the

use of a customized ChatGPT personalized to the

topic of continuity indeed offered significant benefits

compared to the default, non-customized version.

Teachers highlighted the improved relevance,

accuracy, and pedagogical alignment of the resources

generated by the instructed model. The personalized

instructions enabled the chatbot to address specific

learning objectives more effectively, supporting the

creation of tasks and explanations that were directly

aligned with teaching goals.

Second, the instructed ChatGPT positively

influenced all three dimensions of the KTMT

framework:

Task Characteristics: The tool facilitated the

design of tasks that varied in structure and cognitive

complexity, enabling teachers to challenge students at

multiple levels of understanding.

Balance Among Representations: It promoted

transitions between algebraic, graphical, and tabular

formats, fostering an integrated understanding of

continuity.

Experimentation and Justification: It supported

activities that balanced hands-on exploration with

rigorous proof-based reasoning, aligning with the

framework’s emphasis on connecting empirical

exploration with deductive rigor.

These findings highlight the critical role of

teacher preparation in maximizing the potential of AI

tools like ChatGPT. The initial skepticism expressed

by participants—regarding the reliability,

appropriateness, and usability of the technology—

underscores the need for targeted training to build

teachers’ confidence and competence in integrating

AI into their instructional practices. Without this

preparation, there is a risk of reinforcing negative

attitudes toward technological innovation and

limiting the effectiveness of such tools in educational

settings.

In conclusion, the study demonstrates that the

customization of ChatGPT not only enhances its

functionality but also aligns with the pedagogical

priorities outlined in the KTMT framework. By

addressing key dimensions such as task diversity,

representational balance, and the integration of

experimentation with justification, the instructed

ChatGPT provides a powerful example of how AI

tools can transform mathematics education.

CSEDU 2025 - 17th International Conference on Computer Supported Education

410

5 CONCLUSION AND FUTURE

WORKS

The study faced challenges related to technology

adoption and skepticism from participants,

particularly regarding the reliability of ChatGPT and

their ability to use it effectively. These perceptions

influenced how they engaged with the tool, despite

ongoing support. The study also highlighted the need

for more comprehensive training programs to build

teachers’ confidence and competence in using AI

tools. Another challenge was the limited scope of

customization applied to ChatGPT. While

improvements were observed compared to the default

version, further fine-tuning could yield even better

results, emphasizing the importance of iterative

development.

Additionally, while the study provided valuable

qualitative insights, it would benefit from quantitative

analysis to support the findings. A mixed-methods

approach, combining both qualitative and

quantitative data, would strengthen the evaluation of

ChatGPT’s impact on teaching practices, such as

through pre- and post-assessments of teachers’

resource design or students’ learning outcomes.

In conclusion, the study explored the potential of

a customized ChatGPT model to support teaching

continuity for real-valued functions. Using the KTMT

framework, the research showed how personalized AI

tools can align with pedagogical goals, improving

task design, representation balance, and the interplay

between experimentation and justification.

Significant improvements in instructional resource

quality were observed with the customized ChatGPT.

Future work will refine the model, expand the

participant sample, and integrate a mixed-methods

approach to further investigate AI’s role in education,

contributing to the integration of advanced

technologies in mathematics teaching.

ACKNOWLEDGEMENTS

The research of Maria Lucia Bernardi is funded by a

PhD fellowship within the framework of the Italian

“D.M. n. 118/23”- under the National Recovery and

Resilience Plan, Mission 4, Component 1, Investment

4.1 - PhD Project “Tech4Math-Math4STEM” (CUP

H91I23000500007).

REFERENCES

Bernardi, M. L., Capone, R., Faggiano, E., & Rocha, H.

(2024). Exploring Pre-service Mathematics Teachers'

Perceptions of Generative AI in Mathematics

Education: A Pilot Study. In 2nd International

Conference on Math Education and Technology

(ICMET 2024). Book of Abstracts (pp. 90-91).

Bernardi, M.L., Capone R., Faggiano E., Troilo, F. (2024b).

Teachers exploring the potential of generative AI in

mathematics teaching. In Proceedings of the 17th ERME

Topic Conference MEDA4, 89-96

Capone & Faggiano (to appear). Generative artificial

intelligence scaffolding students’ understanding of triple

integrals. In Proceedings of the 15th International

Congress on Mathematical Education. Sydney.

Dasari, D., Hendriyanto, A., Sahara, S., Suryadi, D.,

Muhaimin, L. H., Chao, T., & Fitriana, L. (2024,

January). ChatGPT in didactical tetrahedron, does it

make an exception? A case study in mathematics

teaching and learning. In Frontiers in Education(Vol. 8,

p. 1295413). Frontiers Media SA.

Drijvers, P., & Sinclair, N. (2023). The role of digital

technologies in mathematics education: purposes and

perspectives. ZDM–Mathematics Education, 1-10.

Duval, R. (2006). A cognitive analysis of problems of

comprehension in a learning of mathematics.

Educational Studies in Mathematics, 61(1–2), 103–131

Goos, M., & Bennison, A. (2008). Surveying the technology

landscape: Teachers’ use of technology in secondary

mathematics classrooms. Mathematics Education

Research Journal, 20(3), 102–130

Govender, R. (2023). The impact of artificial intelligence

and the future of ChatGPT for mathematics teaching and

learning in schools and higher education. Pythagoras,

44(1), 1-2.

Lesseig, K. (2016). Investigating mathematical knowledge

for teaching proof in professional development.

International Journal of Research in Education and

Science, 2, 253–270

Mishra, P., & Koehler, M. (2006). Technological

pedagogical content knowledge: A framework for

teacher knowledge. Teachers College Record, 108,

1017–1054.

Pavlova, N. H. (2024). Flipped dialogic learning method

with ChatGPT: A case study. International Electronic

Journal of Mathematics Education, 19(1), em0764.

Ponte, J. (2005). Gestão curricular em Matemática. In GTI

(Eds.), O professor e o desenvolvimento curricular (pp.

11–34). Lisbon: APM.

Rocha, H. (2013). Knowledge for Teaching Mathematics

with Technology—A new framework of teacher

knowledge. In A. Lindmeier & A. Heinze (Eds.),

Proceedings of the 37th PME, vol. 4 (pp. 105–112).

Kiel: PME

Rocha, H. (2020). Using tasks to develop pre-service

teachers’ knowledge for teaching mathematics with

digital technology. ZDM, 1-16.

Shulman, L. (1986). Those who understand: Knowledge

growth in teaching. Educational Researcher, 15(2), 4–

14.

Empowering Mathematics Educators: Integrating ChatGPT as a Tool for Innovative Teaching Practices

411