[Link]

php/AISAI

Volume 1 Issue 1 Year 2026 | Page 1-11 | e-ISSN: xxxx-xxxx | ISSN: xxxx-xxxx
Received: 18-01-2026 | Revised: 27-01-2026 | Accepted: 10-02-2026

The Impact of Moderating ChatGPT Usage on Problem-Solving Skills

among Computer Science Students
Nadhifa Fairuz Salsabila1, Elsa Dewi Rahmalia Sari2, Intan Dwi Safrida3, Melati
Mahardika4, Nur Hayatin5
1,2,3,4,5
Universitas Muhammadiyah Malang, Kota Malang, Indonesia, 65145
E-mail: : nadhifasalsabila@[Link] 1, elsaa@[Link].id2, intandwis26web-
mail@[Link].id3, melatii@[Link].id4, noorhayatin@[Link].id5
*Correspondence: nadhifasalsabila@[Link]

Abstract: The rapid development of Artificial Intelligence (AI) in higher education has encouraged
the adoption of various digital technologies as learning support tools, including ChatGPT. In
Computer Science programs, students are frequently confronted with complex academic
problems, making technological support valuable for enhancing their problem-solving skills. This
study aims to analyze the effect of ChatGPT usage on students’ problem-solving ability and to
examine the role of usage frequency as a moderating variable. This research employs a
quantitative approach involving 59 Computer Science students from Universitas Muhammadiyah
Malang. The data were analyzed using validity testing, reliability testing, correlation analysis,
and regression analysis. The results indicate that ChatGPT usage has a highly significant effect
on students’ problem-solving skills. In addition, usage frequency was found to strengthen the
relationship between ChatGPT usage and problem-solving ability. The study concludes that
ChatGPT can serve as an effective learning support tool when used in a guided and responsible
manner within the higher education learning process.

Keywords: ChatGPT, artificial intelligence, problem-solving, students, Computer Science.

1. Introduction

The development of Artificial Intelligence (AI) in education in Indonesia has grown rapidly,
particularly in higher education learning processes[1]. AI is increasingly used as a learning
support tool that helps students understand course concepts, complete academic assignments, and
improve learning efficiency. One AI application frequently used by students is ChatGPT, which
functions as a text-based system that provides explanations and solutions for various academic
problems[2].
In Computer Science programs, students face relatively complex challenges, such as
understanding algorithms, programming, and debugging code[3]. In this context, ChatGPT can
provide quick and structured access to problem-solving support, helping students comprehend the
steps required to solve a given problem. Nevertheless, several studies have raised concerns about
excessive ChatGPT use, particularly regarding its potential impact on critical thinking and
students’ dependency on AI[4].
Most previous studies have discussed the use of ChatGPT from conceptual perspectives, literature
reviews, or qualitative approaches, and have tended to emphasize the general impact of AI on
learning. However, there is still limited empirical research that quantitatively measures the effect

1
Applied Informatics for Sustainability and Artificial Intelligence Journal, Volume 1, Issue 1, March 2026: xx-xx ◼
ISSN: 1978-1520

of ChatGPT usage on students’ problem-solving skills, especially by considering usage frequency

as a strengthening factor. This condition indicates a research gap that warrants further
investigation[5].
Therefore, this study aims to analyze the effect of ChatGPT usage on the problem-solving ability
of Computer Science students and to examine the moderating role of usage frequency. This
research adopts a quantitative approach involving 59 Computer Science students at Universitas
Muhammadiyah Malang. The findings are expected to contribute to a better understanding of
ChatGPT as an effective and responsible learning support tool in higher education.

2. Research Methods
This study was conducted through several systematically arranged stages to ensure that the data
collection and analysis processes were carried out accurately[6]. The research workflow begins
with problem identification and ends with conclusion drawing. Visually, the research stages are
presented in Figure 1.

Figure 1. Research Flow

2.1 Problem Identification and Literature Review
The first stage involves identifying the research problem by examining the increasing use of
ChatGPT among Computer Science students and its potential impact on problem-solving skills.
An initial literature review was conducted to identify existing findings and determine the research
gap[7].

2
[Link]

2.2 Problem Formulation and Variable Definition

In the second stage, the research objectives and variables were defined. ChatGPT usage was
established as the independent variable, problem-solving ability as the dependent variable, and
usage frequency as the moderating variable[8]. A research instrument in the form of a
questionnaire was then designed to measure both ChatGPT usage intensity and students’
perceptions of their problem-solving abilities. The research variables are defined as follows:
1. Independent Variable (X): ChatGPT usage
2. Dependent Variable (Y): Problem-solving ability
3. Moderating Variable (Z): Frequency or intensity of usage this stage ensures a focused
research direction for measuring the impact of technology usage[9].
2.3 Instrument Development and Research Data
The study uses primary data collected directly from respondents. The research instrument is a
closed-ended questionnaire designed using a five-point Likert scale (1–5) to measure students’
perspectives. The questionnaire assesses ChatGPT usage intensity and indicators of technical
problem-solving skills.
2.4 Data Collection
Data collection was conducted by distributing the questionnaire to 59 Computer Science students
at Universitas Muhammadiyah Malang. The collected responses were then screened to ensure
completeness and suitability for analysis.
2.5 Data Analysis
After data collection, statistical analyses were performed, including validity testing, reliability
testing, correlation analysis, and regression analysis. These analyses were used to examine the
effect of ChatGPT usage on problem-solving ability and to evaluate the role of usage frequency
as a moderating variable. The final stage involved drawing conclusions by interpreting the
statistical results and providing recommendations regarding the responsible use of ChatGPT in
Computer Science education.

3. Results and Discussion

Based on the average-score chart of ChatGPT usage indicators, all indicators show mean values
above 3.50, indicating that overall ChatGPT utilization among Computer Science students falls
into the high category can show in the Figure 2.

Figure 2. Indicator

3
Applied Informatics for Sustainability and Artificial Intelligence Journal, Volume 1, Issue 1, March 2026: xx-xx ◼
ISSN: 1978-1520

Indicator P13 obtained the highest score and reflects a positive contribution to problem-solving.
This suggests that students perceive ChatGPT as beneficial in supporting their overall problem-
solving performance, particularly in analyzing problems and identifying solutions more
effectively. Indicators P1 and P2 also achieved high scores, implying that ChatGPT plays an
important role in the early stages of problem-solving by helping students understand the problem
context and conduct initial analysis before determining a solution.
Indicators P9 and P10 received the same score, indicating that ChatGPT may enhance students’
confidence and support more logical decision-making when solving academic problems.
Meanwhile, P7 recorded the lowest score; however, it remained relatively high, suggesting that
ChatGPT does not fully replace students’ logical thinking, but rather serves as a supportive tool.
Indicators P3 and P4 suggest that ChatGPT is fairly effective as an exploratory tool for seeking
alternative solutions and checking students’ work.
3.1 Analysis of Research Results
Overall, the chart indicates that ChatGPT is used more as a learning support tool than as a means
to obtain instant answers. Students remain actively engaged in learning, while ChatGPT is used
to strengthen understanding, improve learning efficiency, and verify solutions. This finding aligns
with previous studies suggesting that guided and responsible use of ChatGPT can positively affect
students’ problem-solving performance can show in the Figure 3.

Figure 3. Score per Question

In addition, the high scores on problem-solving impact indicators suggest that higher usage
frequency may further strengthen the positive effect. The statistical results also confirm a positive
and significant effect of ChatGPT usage on students’ problem-solving ability:
1. Linear Regression Equation
The regression model is:
𝑌 = 𝑎 + 𝑏𝑋 (1)
Based on the analysis, the constant value is a = 2.220 and the regression coefficient is b =
0.651, resulting in:
𝑌 = 2.220 + 0.651𝑋 (2)
This indicates that each increase in ChatGPT usage corresponds to an increase of 0.651 units
in students’ problem-solving ability[10].

4
[Link]

2 Coefficient of Determination
The analysis shows 𝑅 = 0.769 and 𝑅² = 0.591, meaning that 59.1% of the variance in
students’ problem-solving ability can be explained by ChatGPT usage, while the remaining
40.9% is influenced by other factors not included in this study.
3. t-Test
The t-test result shows 𝑡 = 9.001 with 𝑝 < 0.005, indicating that ChatGPT usage has a
statistically significant effect on students’ problem-solving ability.
4. F-Test
The regression model is significant with 𝑝 < 0.005, indicating that the model is appropriate
for explaining the relationship between ChatGPT usage and problem-solving ability. The
analysis reports 𝐹 = 81.014.
3.2 Discussion
The findings demonstrate that ChatGPT usage has a significant influence on the problem-solving
ability of Computer Science students. Students do not use ChatGPT merely to obtain instant
answers; instead, it is used to support understanding of problem-solving steps, analyze problems,
and verify programming code. Such use can improve learning efficiency and facilitate
comprehension of technically complex material[11].
This study also indicates that higher frequency of ChatGPT usage is associated with greater
improvement in problem-solving ability. Students become more confident in facing problems,
analyzing them, formulating solutions, and making logical decisions. This provides a different
perspective from studies emphasizing the risks of dependency and reduced critical thinking due
to excessive AI use. In this study, usage frequency functions as a moderating factor that
strengthens the positive impact of ChatGPT when used in a guided and responsible manner[12].
Therefore, the results support the view that ChatGPT does not necessarily reduce students’
thinking ability. When used as a learning aid, ChatGPT can contribute positively to students’
problem-solving development, particularly in Computer Science programs that require strong
analytical and problem-solving capabilities.

4. Conclusions
This study concludes that ChatGPT usage has a significant positive effect on improving the
problem-solving ability of Computer Science students. ChatGPT is not only used to obtain instant
answers but also as a learning support tool to understand problem-solving procedures, particularly
for technical and complex topics. The findings also show that usage frequency acts as a
moderating variable that strengthens the relationship between ChatGPT usage and students’
problem-solving ability. Students who use ChatGPT consistently and purposefully tend to
demonstrate better analytical and problem-solving performance than those who use it
infrequently.
Overall, the study confirms that ChatGPT can have a positive impact on students’ thinking
processes and problem-solving performance when used responsibly. These findings may serve as
a basis for educators and higher education institutions to integrate AI technologies as learning
tools that support students’ analytical development. Students are encouraged to use ChatGPT
responsibly and purposefully as a learning aid particularly to understand concepts and solution
steps rather than as a substitute for independent thinking. Future studies are recommended to
involve a larger and more diverse sample and include additional variables to produce more
comprehensive results.

5
Applied Informatics for Sustainability and Artificial Intelligence Journal, Volume 1, Issue 1, March 2026: xx-xx ◼
ISSN: 1978-1520

References

[1] H. Crompton and D. Burke, “Artificial intelligence in higher education: the state of the
field,” International Journal of Educational Technology in Higher Education, vol. 20, no.
1, p. 22, Apr. 2023, doi: 10.1186/s41239-023-00392-8.
[2] N. A. Dahri et al., “Investigating AI-based academic support acceptance and its impact on
students’ performance in Malaysian and Pakistani higher education institutions,” Educ.
Inf. Technol. (Dordr)., vol. 29, no. 14, pp. 18695–18744, Oct. 2024, doi: 10.1007/s10639-
024-12599-x.
[3] J. Babu, “Role of AI Tools in Boosting Student’s Learning Outcome,” ComFin Research,
vol. 13, no. S1-i2, pp. 74–76, Apr. 2025, doi: 10.34293/commerce.v13iS1-i2.8738.
[4] R. Suriano, A. Plebe, A. Acciai, and R. A. Fabio, “Student interaction with ChatGPT can
promote complex critical thinking skills,” Learn. Instr., vol. 95, p. 102011, Feb. 2025, doi:
10.1016/[Link].2024.102011.
[5] M. Urban et al., “ChatGPT improves creative problem-solving performance in university
students: An experimental study,” Comput. Educ., vol. 215, p. 105031, Jul. 2024, doi:
10.1016/[Link].2024.105031.
[6] Y. Wu, S. Gao, S. Zhang, X. Dou, X. Wang, and Q. Li, “From Requirement to Solution:
Unveiling Problem-Driven Design Patterns in Visual Analytics,” IEEE Trans. Vis.
Comput. Graph., vol. 31, no. 10, pp. 7089–7106, Oct. 2025, doi:
10.1109/TVCG.2025.3538768.
[7] J. Liao, L. Zhong, L. Zhe, H. Xu, M. Liu, and T. Xie, “Scaffolding Computational
Thinking With ChatGPT,” IEEE Transactions on Learning Technologies, vol. 17, pp.
1628–1642, 2024, doi: 10.1109/TLT.2024.3392896.
[8] G. Orrù, A. Piarulli, C. Conversano, and A. Gemignani, “Human-like problem-solving
abilities in large language models using ChatGPT,” Front. Artif. Intell., vol. 6, May 2023,
doi: 10.3389/frai.2023.1199350.
[9] M. Urban et al., “ChatGPT improves creative problem-solving performance in university
students: An experimental study,” Comput. Educ., vol. 215, p. 105031, Jul. 2024, doi:
10.1016/[Link].2024.105031.
[10] Y.-A. Wang, Q. Huang, Z. Yao, and Y. Zhang, “On a class of linear regression methods,”
J. Complex., vol. 82, p. 101826, Jun. 2024, doi: 10.1016/[Link].2024.101826.
[11] B. Arkan, Ö. E. Dallı, and B. Varol, “The impact of ChatGPT training in the nursing
process on nursing students’ problem-solving skills, attitudes towards artificial
intelligence, competency, and satisfaction levels: Single-blind randomized controlled
study,” Nurse Educ. Today, vol. 152, p. 106765, Sep. 2025, doi:
10.1016/[Link].2025.106765.
[12] R. Zhou et al., “Exploring <scp>ChatGPT</scp> ‐Facilitated Scaffolding in
Undergraduates’ Mathematical Problem Solving,” J. Comput. Assist. Learn., vol. 41, no.
4, Aug. 2025, doi: 10.1111/jcal.70077.