See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/386450749

Innovations in Introductory Programming Education: The Role of AI with

Google Colab and Gemini

Article in Education Sciences · December 2024

DOI: 10.3390/educsci14121330

CITATIONS READS

0 34

4 authors, including:

Joe Llerena Izquierdo Raquel Ayala-Carabajo

Politecnica Salesiana University 23 PUBLICATIONS 368 CITATIONS
79 PUBLICATIONS 403 CITATIONS
SEE PROFILE
SEE PROFILE

César Miguel Andrade-Martínez

Politecnica Salesiana University
17 PUBLICATIONS 16 CITATIONS

SEE PROFILE

All content following this page was uploaded by César Miguel Andrade-Martínez on 06 December 2024.

The user has requested enhancement of the downloaded file.

Article
Innovations in Introductory Programming Education: The Role
of AI with Google Colab and Gemini
Joe Llerena-Izquierdo * , Johan Mendez-Reyes , Raquel Ayala-Carabajo and Cesar Andrade-Martinez

GIEACI Research Group, Universidad Politécnica Salesiana, Guayaquil 090101, Ecuador;

[email protected] (J.M.-R.); [email protected] (R.A.-C.); [email protected] (C.A.-M.)
* Correspondence: [email protected]

Abstract: This study explores the impact of artificial intelligence on the teaching of programming,
focusing on the GenAI Gemini tool in Google Colab. It evaluates how this technology influences the
comprehension of fundamental concepts, teaching processes, and effective teaching practices. In this
research, students’ motivation, interest, and satisfaction are determined, as well as the fulfillment and
surpassing of their learning expectations. With a quantitative approach and a quasi-experimental
design, an investigation was carried out in seven programming groups in a polytechnic university
in Guayaquil, Ecuador. The results reveal that the use of GenAI significantly increases interest in
programming, with 91% of the respondents expressing increased enthusiasm. In addition, 90%
feel that the integration of GenAI meets their expectations, and 91% feel that it has exceeded those
expectations in terms of educational support. This study evidences the value of integrating advanced
technologies into education, suggesting that GenAI can transform the teaching of programming.
However, successful implementation depends on timely training of educators, ethics training for
students, ongoing interest in the technology, and a curriculum design that maximizes the capabilities
of GenAI.

Keywords: AI in education; higher education; generative AI; AI-assisted learning; AI ethics; Google
Colab; Gemini
Citation: Llerena-Izquierdo, J.;
Mendez-Reyes, J.; Ayala-Carabajo, R.;
Andrade-Martinez, C. Innovations in
Introductory Programming Education: 1. Introduction
The Role of AI with Google Colab and
In a world where artificial intelligence (AI) technologies are making inroads, traditional
Gemini. Educ. Sci. 2024, 14, 1330.
educational barriers are being challenged as never before [1,2]. This transformation places
https://doi.org/10.3390/
the student at the center of the learning process where the role of the professor is also
educsci14121330
redefined [3,4]. The integration of technologies such as generative artificial intelligence
Academic Editors: Ilya Levin and (GenAI) in higher education promotes AI-assisted learning that optimizes teaching but, at
Alexei Semenov the same time, raises important ethical questions [5,6].
Received: 30 October 2024
On a more specific level, and referring to the training of engineering students, the
Revised: 25 November 2024
understanding of programming concepts at an early introductory stage of university
Accepted: 2 December 2024 studies is essential for academic success, as it lays the foundation for learning more com-
Published: 4 December 2024 plex content [7–10]. In the development of this learning, however, many young people
encounter difficulties in mentally operating with abstract ideas, such as, for example,
instructions for designing control or repetition structures, as well as the storage of data
in vectors and matrices [11]. In this context, the integration of GenAI in the classroom
Copyright: © 2024 by the authors. offers an effective response to overcome these barriers, providing personalized support
Licensee MDPI, Basel, Switzerland.
and algorithmic resources that facilitate the assimilation of these concepts [12–14].
This article is an open access article
In this sense, this study explores how GenAI tools, such as Gemini in a specific context,
distributed under the terms and
can enrich the understanding of programming and the implications this has for teaching
conditions of the Creative Commons
and learning [6,15]. The incorporation of GenAI into educational curricular content can
Attribution (CC BY) license (https://
transform the way in which various disciplines, including programming, are taught and
creativecommons.org/licenses/by/
learned [16]. Platforms such as Google Colab in combination with Gemini facilitate access
4.0/).

Educ. Sci. 2024, 14, 1330. https://doi.org/10.3390/educsci14121330 https://www.mdpi.com/journal/education

Educ. Sci. 2024, 14, 1330 2 of 19

to learning resources, and create interactive environments that foster dynamic and adaptive
learning [17,18].
In this way, motivation and interest are fostered, which are fundamental for the stu-
dents’ commitment and academic success; enabling learning strategies that favor individual
and group performance [19]. This analysis also aims to determine whether the use of GenAI
tools manages to enhance these aspects, as well as to examine whether students’ learning
expectations are met or even exceeded [5,20,21].
The aim of this work is to evaluate the impact of generative artificial intelligence,
specifically Gemini in Google Colab, in the teaching of programming and to contribute to
the literature in specific scenarios. It analyzes how these tools influence the comprehension
of fundamental programming concepts, in complete teaching processes and effective teach-
ing practices where motivation, interest, and satisfaction of students, as well as in meeting
and exceeding their learning expectations are evidenced, and favor educational develop-
ment. Through classroom experience and student perceptions, we also seek to provide a
comprehensive view of the impact that GenAI has on the teaching of programming, with
the objective of identifying effective educational practices focused on the current needs
of students [22]. To determine the impact of using GenAI tools in teaching programming,
research questions are established in three relevant areas (see Table 1).

Table 1. Research questions for the study.

Research Question Area

Comprehension of
RQ1: how does the use of generative artificial intelligence tools, such as
the content
Gemini in Google Colab, affect the comprehension of fundamental program-
ming concepts among students?
Comprehension of
RQ2: to what extent do artificial intelligence tools facilitate the learning of
the content
complex subjects compared to traditional teaching methods?
RQ3: how does the integration of artificial intelligence in the classroom Interest level
influence students’ interest in programming?
RQ4: what is the relationship between the use of artificial intelligence tools Interest level
and students’ motivation to learn programming?
Learning
RQ5: do you meet students’ learning expectations when using artificial
expectations
intelligence in your programming classes, and to what extent do you exceed
these expectations?
Learning
RQ6: what are students’ perceptions of the support they receive for their
expectations
programming learning through artificial intelligence tools?

The first question focuses on the use of these tools that are integrated into the pro-
gramming course, and improves the comprehension of basic concepts, such as the use of
variables, control structures, selection, or repetition, as well as compound structures such as
vectors and arrays. The purpose is to evaluate whether GenAI (Gemini) provides tangible
support in the assimilation of these essential fundamentals [11]. The second question
focuses on the ease of learning complex topics, exploring whether GenAI helps students to
cope with and understand more advanced content for their academic development [6]. The
third and fourth questions address interest in using GenAI tools, and how they influence
motivation in the classroom [11]. The fifth and sixth questions focus on expectations about
student learning when using GenAI compared to traditional methods [23]. This is funda-
mental to understanding whether the integration of advanced technologies, platforms, and
intelligent environments can make learning more engaging and motivating [7]. In addition,
it evaluates the fulfillment and exceeding of expectations in relation to student satisfaction
with the use of GenAI with Google Colab, in a specific scenario and over time, opening
spaces to contribute with new research and add to the recent ones [24].
Educ. Sci. 2024, 14, 1330 3 of 19

2. State of the Art

The study by Perezchica-Vega et al. [25] investigated professors’ experience with
generative artificial intelligence (GenAI) in education. Using a quantitative methodology,
concerns about academic honesty are identified, and the benefits of GenAI for learning, such
as data analysis and the creation of didactic materials, were highlighted. It is concluded
that its integration could improve educational quality, although adjustments in evaluation
mechanisms are still required. Also, in Hernández González et al. [26], the perceptions of
university students about Generative Artificial Intelligence (GenAI) in education were ana-
lyzed. Through a qualitative and descriptive approach, opinions about its advantages and
disadvantages were identified. Most students accepted its use, highlighting improvements
in their learning, although they expressed concerns about ethics and privacy in the use
of this technology. For their part, Padilla Piernas and Martín-García [27] investigated the
perceptions of Spanish university professors about generative artificial intelligence (GenAI)
in higher education, using the AETGE/GATE model. At the same time, the importance
of training programs for an effective implementation of GenAI in learning is emphasized.
The importance of training programs for effective implementation of GenAI in learning
is emphasized, while Solano Hilario et al. [28], in their research, analyzed how the use of
Generative Artificial Intelligence (GenAI) tools improve student learning compared to the
use of traditional tools. Through a systematic review of the literature, the results indicate
an equal preference between both styles, highlighting personalization and feedback as key
factors to optimize learning.
On the other hand, the research conducted by García Peñalvo et al. [29] analyzed the
impact of generative artificial intelligence, especially ChatGPT (2022 and 2023), on educa-
tion. Through a systematic review of the literature, the advantages and disadvantages of its
use in learning are identified. Although it presents a potential for improving educational
processes, it also poses risks and limitations that require careful evaluation and an adequate
response in teaching practice. In addition, Schiavo et al. [30] consider the use of the Technol-
ogy Acceptance Model (TAM) to understand usage intention, individual perception, and
adaptation in the age of AI literacy. In this regard, they identify factors that influence human
behavior patterns, such as perceived usefulness, perceived ease of use, perceived security,
perceived trust, perceived confidence, respect for privacy, perceived anxiety, perceived
social influence, and perceived compatibility. Thus, these data help to facilitate behavioral
patterns, improve educational inclusion, and support the development of the teaching and
learning process. According to Solomovich and Abraham [31], the Technology Acceptance
Model (TAM) contributes to the collection of data that provides insight into behavioral
traits related to trust levels in chatbot users. These empirical results offer the opportunity to
further improve these technology platforms for the benefit of all stakeholders. A necessary
aspect to highlight is that of intrinsic motivation Verdugo-Castro et al. [15].
Likewise, Henrique-Sanches et al. [32], in their quasi-experimental research, studied
the impact of hands-on activities in the Skill and Simulation Laboratory after the pandemic.
Although motivation levels did not change significantly, the high intrinsic motivation and
regulation identified suggest that active methodologies are key to effective learning in
medical students. In the field of robotics, also based on AI in educational applications,
Vitale and Dello Iacono [33], through explanatory research, studied the use of social robots
as an inclusive educational technology to improve mathematics learning. The research
showed that the Pepper robot significantly increased student engagement by offering
personalized support. Its ability to create a stimulating learning environment tailored to
diverse needs was highlighted.
Some studies, such as the one by Rossettini et al. [34], evaluated the accuracy of AI
chatbot responses (ChatGPT-4, Microsoft Copilot, and Google Gemini) in health science
input tests. The intelligent assistants or chatbots performed well overall, although with dif-
ferences in accuracy and narrative coherence. In the study by Fabijan et al. [35], the ability
of AI models to interpret MRI video sequences was analyzed to identify and analyze pedi-
atric brain tumors. The models failed to accurately identify lesions, highlighting the need
Educ. Sci. 2024, 14, 1330 4 of 19

for better targeted integration of AI into medical diagnostics. While Rossettini et al. [34]
evaluated the accuracy and narrative coherence of AI chatbots when using standardized
exam questions in health sciences, Fabijan et al. [35] explored the limited capabilities of AI
models (ChatGPT and Gemini Pro) when analyzing medical MRI videos, i.e., the need to
improve the experience of constructive use of AI tools in medical domain activities requires
incorporating reliable meaningful experiences that enhance the understanding of content
for an AI.
Students in higher education show significant interest in AI chatbots, driven by their
ability to provide immediate feedback, assist with academic tasks, and boost motivation.
However, concerns regarding the reliability of AI responses have also been noted, as these
tools sometimes produce inaccurate or inconsistent information. As Schei et al. [36] points
out, the appeal of AI chatbots lies in their immediate usefulness and accessibility, making
them particularly attractive for students seeking quick solutions to academic challenges.
In a similar vein, Alnasib and Alharbi [37] highlights that the integration of AI tools
like Gemini into the teaching of English as a foreign language generated student inter-
est, although several limitations were identified, such as repetitive responses and issues
with accuracy. While these tools were shown to enhance motivation, their practical effec-
tiveness in fostering language learning remains limited. This suggests that, despite their
potential, improvements in AI response quality are necessary to optimize their role in
educational contexts.
The study by Zichar and Papp [38] explored students’ expectations when using AI tools
for 3D modeling and programming tasks. While these tools offer valuable initial support in
code generation, students are not yet able to fully delegate these tasks to AI. Torres-Peña
et al. [39], in their paper, discussed how AI tools such as ChatGPT and Wolfram Alpha have
raised students’ expectations in mathematics, particularly in calculus, by helping them
solve derivative problems with greater accuracy and conceptual understanding. Conversely,
Yoseph et al. [40] revealed that while spine surgery patients prioritize clear answers, their
learning expectations also focus on a comprehensive understanding of complex medical
procedures, such as those facilitated by AI. This is evidence of how patients can benefit
from the clarity and conciseness that AI can provide when explaining intricate topics.
Similarly, Almassaad et al. [41] describes how the use of GenAI tools in higher educa-
tion in Saudi Arabia has raised expectations of greater efficiency and academic support.
However, students also express concerns about the accuracy of the information provided
by these tools. Taken together, these studies shed light on how AI tools are shaping ex-
pectations in various domains, including education and healthcare, while emphasizing
challenges related to the reliability and practical use of these technologies.

2.1. A Blended Constructivist and Project-Based Learning (PBL) Approach with the Use
of Artificial Intelligence Tools in Education
Constructivist learning theories, articulated by Bruner, Vygotsky, and Piaget, empha-
size that knowledge is actively constructed in the mind of the learner through interactive
and meaningful experiences. These principles are exemplified in recent studies that inte-
grate AI tools into educational practices. Rossettini et al. [34], for example, investigated
the use of AI chatbots for exam assessments, framing it as an educational project in which
students interact with technology to assess their performance in specific contexts. This
approach allows students to reflect on the problem-solving capabilities of AI, in line
with the constructivist view that interaction with tools and feedback are essential for
knowledge construction.
Similarly, Alnasib and Alharbi [37] explored how the AI tool Gemini supports English
language learners, highlighting how students actively engage with linguistic challenges.
By providing immediate feedback and facilitating practice, Gemini helps learners respond
to existing uncertainties, encouraging critical reflection, an integral component of construc-
tivism. Zichar and Papp [38] extends this idea to complex tasks such as 3D modeling and
programming, where students use AI tools to address challenges in coding and design. In
Educ. Sci. 2024, 14, 1330 5 of 19

these tasks, AI serves as a resource and collaborator, promoting deeper engagement with
the technology as students integrate it into problem-solving processes. This aligns with the
principles of project-based learning (PBL), where students address real-world problems by
applying and reflecting on their knowledge.
In the field of mathematics, Torres-Peña et al. [39] examined how AI tools, such
as ChatGPT and Wolfram Alpha, improve the teaching of calculus. These tools enable
students to solve derived problems more accurately, providing immediate feedback that
encourages autonomous learning. The use of AI in this context supports the development
of mathematical understanding, and reinforces constructivist values by allowing students
to actively interact with the material, verify solutions, and reflect on their learning process.
Similarly, Yoseph et al. [40] demonstrates the role of AI in medical education, specifically in
providing clear and understandable explanations of complex procedures, such as cervical
discectomy surgeries. By interacting with AI-generated responses, patients and students
develop a more interactive understanding of medical concepts, which encourages informed
decision-making and reflective learning.
Almassaad et al. [41] highlighted the integration of AI tools in higher education
in Saudi Arabia, focusing on language learning. This study highlights the use of AI as
a facilitator of project-based learning, where students work on specific tasks, such as
improving language proficiency, while reflecting on their progress and adjusting their
learning strategies. Incorporating AI tools into these projects allows students to solve
problems in practical, real-world contexts, reflecting key features of project-based learning
and constructivist learning environments.
The use of AI in education allows students to face authentic and challenging tasks
that require an integration of cognitive and technical skills, fostering autonomous and
collaborative learning, which are characteristics of a blended (constructivist and PBL)
framework. In addition, students interact with AI technologies, allowing them to build
knowledge through practice, problem solving, and continuous feedback (see Table 2).

Table 2. Work related to the areas of study.

Scope References
Rane et al. [6], Imran and Almusharraf [7], Solano Hilario et al.
Comprehension of the content
[28], Rossettini et al. [34], Fabijan et al. [35]
Edwards et al. [8], Hernández González et al. [26], Padilla
Interest level Piernas and Martín-García [27], Schei et al. [36], Alnasib and
Alharbi [37]
Vitale and Dello Iacono [33], Zichar and Papp [38], Torres-Peña
Learning expectations
et al. [39], Yoseph et al. [40], Almassaad et al. [41]

2.2. A Comparative Analysis of Gemini’s Effectiveness Against Other Generative AIs

In recent years, artificial intelligence (AI) models have evolved from being simple
support tools to become powerful enablers of learning, especially in educational and
medical contexts. Among the most prominent models is Gemini, which has proven to
be an effective tool in generating understandable and accessible content for a wide range
of users. Unlike other models such as ChatGPT, which is distinguished by its depth of
understanding and its ability to generate accurate and detailed answers in specialized
fields, Gemini stands out for its clarity and readability, making it particularly valuable in
scenarios where readability and accessibility are crucial. This was evidenced in studies
such as that of Hanci et al. [42], who observed that Gemini produced easier to understand
answers compared to other models in the context of palliative care, although improvements
were required in the quality and depth of the content.
Educ. Sci. 2024, 14, 1330 6 of 19

Comparative analysis between Gemini and other AI models, such as ChatGPT and
Bard, has yielded interesting results in terms of their effectiveness in content comprehension.
In studies conducted by Karaca [43] and Gomez-Cabello et al. [44], it was observed that
Gemini is able to generate content suitable for various educational levels, showing an
outstanding ability to adapt its language and complexity according to the user’s needs.
However, in terms of depth and accuracy in technical areas, ChatGPT outperformed
Gemini, as seen in research on interpreting medical results and decision-making in plastic
surgery. Despite these differences, both models show significant potential for improving
the accessibility of knowledge, especially in contexts where users do not have a deep
understanding of the subject matter.
The study of Gemini’s effectiveness also highlights its potential as a key tool for im-
proving learning expectations. In its comparison with ChatGPT-4 and other models in
educational tasks, Gemini showed superior performance in terms of content appropriate-
ness at lower or simple educational levels. This is particularly relevant when analyzing
the quality of the generated texts in terms of readability. According to Karaca [43], AI
models such as Gemini are able to adapt the difficulty level of texts to the needs of the
audience, making this AI suitable for a diverse range of users, from primary school stu-
dents to university students. In the context of education, this ability to generate accessible
content facilitates learning and improves understanding of complex concepts, positioning
Gemini as a potential ally in the teaching of disciplines that require a clear and direct
pedagogical approach.
However, Gemini’s effectiveness as a learning enhancer must also be evaluated in
terms of its performance in solving specialized questions and the ability to generate accurate
answers in complex areas. While Gemini is effective at generating readable and understand-
able answers, its performance on more complex tasks, such as interpreting medical results
or solving advanced chemistry problems, often falls short of the level of accuracy and depth
demonstrated by ChatGPT-4. This point was highlighted by Kharchenko and Babenko [45],
who observed that, while Gemini is proficient in tasks that do not require deep logical
reasoning, in complex tasks related to science and medicine, ChatGPT showed a greater
ability to handle abstract concepts and generate more complete solutions. In this sense,
although Gemini represents a significant advance in the generation of accessible content,
its role as an educational and support tool remains complementary to more specialized
models, such as ChatGPT.
Gemini has an advantage in the readability of answers, especially compared to
ChatGPT-4, which tends to generate more complex answers. However, in terms of ac-
curacy and quality of answers, especially in specialized contexts such as medicine or
chemistry, Gemini was outperformed by ChatGPT in most of the studies analyzed (see
Table 3).

Table 3. Works relating Gemini to the fields of study.

Scope Relationship Reference

Gemini shows significant differences in
quality, readability, clarity, and precision in
Comprehension Hanci et al. [42], Karaca [43],
specific fields, i.e., it is more concise and
of the content Sonmezoglu and Sonmezoglu
excels in the readability of texts.
[46]
Gemini shows competence in providing quick
and direct answers that capture the user’s
Interest level Kharchenko and Babenko [45],
attention, but still lacks the ability to generate
Meyer et al. [47], Durmaz En-
deeper interest in more complex topics.
gin et al. [48]
Gemini performs competitively, shows a
better fit in terms of readability at simpler
Learning Gomez-Cabello et al. [44],
educational levels and limitations in tasks
expectations Farghal and Haider [49], Is
requiring deep logical reasoning.
and Menekseoglu [50]
Educ. Sci. 2024, 14, 1330 7 of 19

In summary, the use of generative artificial intelligence, specifically Gemini in Google

Colab, influences university students’ understanding and interest in learning programming,
from basic to advanced levels, which implicitly aligns with these approaches by promoting
active learning, interaction with technology, and problem solving.

3. Materials and Methods

This study implements a research methodology with a quantitative approach and a
quasi-experimental design, which falls within an empirical-analytical framework. Seven
courses of programming, a subject with high enrollment and belonging to the first level of
studies in engineering degrees of a polytechnic university in the city of Guayaquil, Ecuador,
took part. A work strategy was applied based on laboratory activities in six phases in each
two-hour laboratory day, twice a week, for five months. In addition, the survey technique
is applied to a population of 250 students at the end of the five months, at the end of the
academic period. With a confidence level of 95% and a margin of error of 5%, a set of
six structured questions with Likert scale are established, and focus on important aspects
that allow us to evaluate how the GenAI tool, such as Gemini in Google Colab, affect the
learning experience of students.
The six phases of work as shown in Figure 1 are established on the basis of activities
planned by the team of professors, known as teaching staff, who run these courses [51].
The cloister of professors are professionals, professors, and researchers who share in their
teaching activity the development of the subject of programming [52,53]. This group
of professors is responsible for designing, developing, implementing, and evaluating the
activities of the subject syllabus [54,55]. From an instructional design applied in a basic class-
room [56], professors integrated research-based activities in an educational environment,
based on their didactic experience and techno-pedagogical skills [57], as well as curricular
adaptations based on the results of the evaluations of the course activities [11,58–60].

Figure 1. The professor presents a problem to be solved algorithmically and its corresponding
analysis (a), students use Google Colab for the implementation of their proposal (b), students develop
a prototype without the use of Gemini (c), then refine their work online together with their peers (d),
the professor evaluates their work (e), and the generative artificial intelligence Gemini presents a
different proposal (f).

3.1. Working Phase 1

In this phase, the course’s professor presents a problem to be solved algorithmically,
including the use of simple and compound variables, as well as the control, selection, and
repetition structures found in the syllabus, as an introduction to specific topics. With this,
he begins the subject by presenting a problem, applies the technique of algorithms by means
of pseudocode and its corresponding flowchart, and then its implementation in Python.
Thus, the explanation, dialogue, and discourse are directed to its corresponding analysis
from the accompaniment of schemes, such as pseudocode and its corresponding flowchart,
to finally implement it [37]. In this phase, the use of programs such as PseInt (available at
https://pseint.sourceforge.net/, accessed on 30 November 2024), among others, allows
students to become familiar with a process where algorithmic reflection and analysis are
required for translation into programming language (see Figure 1a).
Educ. Sci. 2024, 14, 1330 8 of 19

3.2. Working Phase 2

In this phase, the course professor requests the use of the Google Colab (Available at
https://colab.research.google.com/, accessed on 30 November 2024) development environ-
ment for a variety of factors. Among these factors is the use of Gmail accounts which, in the
context of this study, all students use on their Android mobile devices. With this, under this
particular context, it encourages mobility, quick access, automatic synchronization, the use
of default tools, and an optimized interface, allowing students to collaborate in real time,
an environment that facilitates their adaptability, operability, access to the files created,
and their backup and saving in the Google Drive cloud, which allows for a convenient
and efficient experience. They are also introduced to Gemini’s built-in assistant [34] (see
Figure 1b).

3.3. Working Phase 3

In this phase, the students use the teams for individual and personal work on the
development of a solution. In other words, the analysis developed is implemented in
the Google Colab environment. With this, reflection, reasoning, logic, memorization, and
recall of what has been learned from the theoretical aspect converge in this space [41]. The
resources used by the students are those documents produced by the tutor professors,
and hosted in the virtual environment of the course, in a Moodle (available at https:
//moodle.org/, access on 30 November 2024) work environment. In other words, having
resources in a virtual classroom that integrates external technologies, such as Google
applications, allows for an ideal working ecosystem for the subject of programming (see
Figure 1c).

3.4. Working Phase 4

In this phase, students establish a collaborative workspace with the professor. The
professor shares the link of his proposal with the students for a collaborative access work.
With this, students can develop two situations. The first situation is to work together
with the professor in their environment, propose changes to their code, exchange lines
of instruction, and test on the professor’s proposal combined with any student [51]. The
second situation is that the student can improve their proposal from the collaborative work
of all, in the company of the professor and their fellow peers, knowing and applying good
programming practices (see Figure 1d).

3.5. Working Phase 5

In this phase, students share and present their working prototype so that the professor
and their peers can improve the lines of instruction. This collaborative work is developed
among assigned peers, so that each code can be reviewed and improved. In this way,
students become peer reviewers in an environment where everyone can contribute to the
changes [54]. The importance of having additional projection equipment, as well as remote
internet access using desktop computers and mobile devices, allows the professor to have
movement within the work lab, generating a close and, at the same time, a mobile space in
this scenario that combines the traditional with the experience of cloud-based tools (see
Figure 1e).

3.6. Working Phase 6

In this phase, the professor and the students of the course begin their interaction with
GenAI Gemini, starting with small queries [38]. These queries are first carried out by the
professor and projected to the whole course. In this way, the student begins to recognize
the style of ‘Prompt’ to design or perform in order to train or guide Gemini during the
refinement of their proposal. This space becomes a question and answer experience with the
conviction to test and improve. In many cases, the professor tests the answers, and instructs
Gemini to rewrite the code, or not to use complexity but simplicity in the lines. In this
scenario, the possibilities of obtaining short lines, use of unstudied functions, complexity
Educ. Sci. 2024, 14, 1330 9 of 19

in syntax for first year students may provoke a preoccupation with learning what the
AI knows rather than a motivation to learn to program [60]. Finally, the solution to be
obtained, from the initial approach, is chosen by the group after establishing the conditions
of the requested problem in the elaboration of the code, which is a simple code for its
understanding, avoiding codes with complex syntax (see Figure 1f).

4. Results
The survey applied electronically, using the Google Forms tool, included six questions
that addressed different sub-fields of study, and were structured on a five-level Likert scale
of ‘Strongly Disagree’, ‘Disagree’, ‘Neither Agree nor Disagree’, ‘Agree’, and ‘Strongly
Agree’. Table 4 presents the relationship between the variables determined and the ques-
tions of the questionnaire in the evaluation of the impact of the use of generative artificial
intelligence, Gemini, in the subject of programming in Google Colab.

Table 4. Relationships between variables and questions in the evaluation of the impact of the use of
GenAI Gemini in the subject of programming.

Sub-Area Variable Questionnaire Question

Understanding program- Level of understanding of ba- Q1: has the use of GenAI tools
ming concepts sic concepts (e.g., arrays, con- such as Gemini in Google Colab
trol structures) using GenAI. improved my understanding of
programming concepts?
Ease of learning complex Perceived ease of understand- Q2: do I find it easier to under-
issues ing complex issues when us- stand complex topics, such as
ing GenAI tools matrices, when I use GenAI in
programming classes?
Interest in the subject Level of interest in learning Q3: has the use of GenAI in pro-
matter programming using GenAI gramming classes increased my
compared to traditional meth- interest in learning more about
ods the subject?
Motivation for learning Perceived level of motivation Q4: do I feel more motivated to
programming to use GenAI tools learn programming when using
GenAI tools compared to tradi-
tional methods?
Fulfilling expectations Satisfaction regarding the ful- Q5: does the integration of
fillment of learning expecta- GenAI in my programming
tions with GenAI classes meet my expectations of
understanding the concepts ef-
fectively?
Exceeding expectations Perception of support received Q6: do I feel that the use of
for learning programming GenAI has exceeded my expec-
through GenAI tations in terms of support for
learning programming?

Table 4 offers an approach that integrates the relationship between the assessed sub-
domains, the relevant variables, and questionnaire questions in the context of the use
of GenAI tools in programming education [61]. It addresses the understanding of pro-
gramming concepts, where the level of understanding of basic notions, syntax, control
structures, and data storage with the use of arrays and matrices is evaluated, facilitated by
the experience with Gemini GenAI. Through the question formulated in the first sub-area,
we seek to determine whether tools such as the Gemini GenAI in Google Colab really
contribute to improving the understanding of these fundamental concepts [62]. The level of
understanding is fundamental to determine a solid foundation in programming fundamen-
tals and is essential for academic success. Another relevant aspect is the ease of learning
complex topics, which focuses on students’ perception of their ability to understand ad-
vanced content with the support of the GenAI. The associated question invites participants
Educ. Sci. 2024, 14, 1330 10 of 19

to reflect on whether the GenAI facilitates the understanding of difficult topics, such as the
use of vectors and matrices for data storage, during lectures. This highlights the potential
of GenAI to simplify the learning of complex content [14].
Interest in the subject of programming is also assessed, focusing on the level of interest
students show in learning programming through the use and support of a GenAI compared
to traditional methods. This question shows how the integration of GenAI can foster greater
interest in learning, which is a key means of academic engagement. In relation to motivation
to learn programming, we seek to understand whether students feel more motivated when
using GenAI tools. The question explores whether this technology, such as the use of GenAI
tools, improves motivation compared to conventional teaching approaches. Motivation is a
decisive factor influencing dedication and commitment to learning.
It also examines the fulfillment of expectations by analyzing students’ satisfaction
with how the GenAI meets their learning expectations. The question posed focuses on
whether the integration of GenAI in programming classes, in conjunction with the professor,
enables students to understand the concepts effectively. Satisfaction in the learning process
is important for assessing the quality of teaching. Exceeding expectations is considered by
assessing how students perceive the support received for their learning through the GenAI.
This question asks whether the GenAI has exceeded expectations in terms of educational
support. This aspect assesses the fulfillment of expectations, and highlights the additional
impact the GenAI can have on the educational experience.

4.1. Analysis of the Survey Responses

The results generally show a positive trend towards the integration of GenAI tools
for learning the subject of programming. Most students report significant improvements
in their understanding, motivation, and interest, highlighting the potential of GenAI to
transform programming education. These findings highlight the need to continue to explore
and apply GenAI technologies in educational settings to foster more dynamic and effective
learning (see Figure 2).

Figure 2. Total survey results presented in percentages.

For question Q1 of the survey, 55% of the students strongly agreed that the use of
GenAI has improved their understanding of basic concepts, 32% agreed, and 12% neither
agreed nor disagreed, while only 1% disagreed. This indicates that 87% of the participants
agree, indicating a strong acceptance of the effectiveness of the GenAI Gemini in Google
Colab in learning programming fundamentals (see Figure 3).
Educ. Sci. 2024, 14, 1330 11 of 19

Figure 3. Total results of question 1 of the survey in percentages.

For question Q2 of the survey, 47% of students said they strongly agreed that GenAI
has made it easier for them to understand complex topics, with a further 42% saying they
agreed. A further 9% neither agreed nor disagreed, while a further 2% disagreed. This
result shows that the GenAI Gemini in Google Colab not only helps with basic concepts,
but is also useful for more advanced topics (see Figure 4).

Figure 4. Total results of question 2 of the survey in percentages.

For question Q3 of the survey, 51% of students strongly agree that the use of GenAI
Gemini in Google Colab has increased their interest in learning more about programming.
A further 40% agreed. A further 8% neither agreed nor disagreed, while 1% disagreed.
This finding is critical, as greater interest is often correlated with better participation and
engagement in learning (see Figure 5).

Figure 5. Total results of question 3 of the survey in percentages.

For question Q4 of the survey, 46% of students strongly agreed that they feel more
motivated to learn programming because of the GenAI Gemini tools in Google Colab. A
further 40% agreed. A further 10% neither agreed nor disagreed, while 4% disagreed. This
increase in motivation is essential to maintain interest and perseverance in learning (see
Figure 6).

Figure 6. Total results of question 4 of the survey in percentages.

Educ. Sci. 2024, 14, 1330 12 of 19

For survey question Q5, 46% of students strongly agreed that the integration of the
GenAI Gemini with Google Colab meets their learning expectations, indicating that, overall,
expectations are being met. A further 44% agreed, and 9% neither agreed nor disagreed,
while 1% disagreed, suggesting that there is still room for improvement in satisfaction (see
Figure 7).

Figure 7. Total results of question 5 of the survey in percentages.

For question Q6 of the survey, 49% of students felt that they strongly agreed that the use
of GenAI has exceeded their expectations in terms of learning support. This is encouraging,
as it implies that the GenAI Gemini in Google Colab is not only meeting expectations, but
is also providing additional value in the educational process. An additional 42% agreed. A
further 8% neither agreed nor disagreed, while 1% disagreed (see Figure 8).

Figure 8. Total results for question 6 of the survey in percentages.

The results show a positive trend towards the integration of GenAI tools in program-
ming education. Most students report significant improvements in their understanding,
motivation, and interest, highlighting the potential of GenAI to transform programming
education. These findings point to the need for further research in exploring and ap-
plying GenAI technologies in specific educational settings to foster more dynamic and
effective learning.
Finally, the results of the number of students who passed the subject in percentages
of the period from October 2023 to March 2024 compared to the period from April to
September 2024, show an improvement of 1%, using strategies that involve intelligent
tools, such as the integration of the GenAI Gemini, which benefit students in their learning
process and professors in their continuous monitoring or follow-up. That is, for the study
period where workspaces for code generation were integrated, the perception of students
has been of greater importance to have an intelligent assistant in conjunction with the
professor, establishing a way of working in the classroom for analysis and implementation
supported with a GenAI tool [58–60] (see Figure 9).

Figure 9. Percentages of students approved and not approved.

Educ. Sci. 2024, 14, 1330 13 of 19

4.2. Statistical Analysis

The results of the correlation analysis reveal significant relationships between the
variables. Pearson’s and Spearman’s correlation coefficients, together with their respective
p-values, are presented in Table 5.

Table 5. Correlation table.

Variable Q1 Q2 Q3 Q4 Q5 Q6
Q1 Pearson’s r –
p-value –
Spearman’s rho –
p-value –
Kendall’s Tau B –
p-value –
Q2 Pearson’s r 0.598 –
p-value <0.001 –
Spearman’s rho 0.633 –
p-value <0.001 –
Kendall’s Tau B 0.601 –
p-value <0.001 –
Q3 Pearson’s r 0.579 0.619 –
p-value <0.001 <0.001 –
Spearman’s rho 0.574 0.628 –
p-value <0.001 <0.001 –
Kendall’s Tau B 0.540 0.595 –
p-value <0.001 <0.001 –
Q4 Pearson’s r 0.616 0.653 0.672 –
p-value <0.001 <0.001 <0.001 –
Spearman’s rho 0.611 0.693 0.696 –
p-value <0.001 <0.001 <0.001 –
Kendall’s Tau B 0.580 0.657 0.660 –
p-value <0.001 <0.001 <0.001 –
Q5 Pearson’s r 0.532 0.457 0.664 0.596 –
p-value <0.001 <0.001 <0.001 <0.001 –
Spearman’s rho 0.581 0.542 0.691 0.659 –
p-value <0.001 <0.001 <0.001 <0.001 –
Kendall’s Tau B 0.546 0.509 0.665 0.616 –
p-value <0.001 <0.001 <0.001 <0.001 –
Q6 Pearson’s r 0.602 0.586 0.681 0.766 0.614 –
p-value <0.001 <0.001 <0.001 <0.001 <0.001 –
Spearman’s rho 0.604 0.641 0.700 0.782 0.699 –
p-value <0.001 <0.001 <0.001 <0.001 <0.001 –
Kendall’s Tau B 0.575 0.604 0.669 0.748 0.664 –
p-value <0.001 <0.001 <0.001 <0.001 <0.001 –
Educ. Sci. 2024, 14, 1330 14 of 19

In the context of using generative artificial intelligence tools in teaching programming,

a high and positive Pearson’s r (e.g., 0.6 or higher) between variables, such as concept
understanding and motivation, is evidence that as students feel a greater understanding of
programming concepts, they also tend to feel more motivated to learn.
A low p-value associated with Pearson’s r (typically less than 0.05) indicates that
there is a high probability that the observed correlation is not due to chance, suggesting
a significant relationship between the variables evaluated. In the context of correlation
analysis, the p-value allows us to determine whether the relationship between the variables
is statistically significant or whether it could be attributable to chance. For example, if the
p-value associated with a Pearson correlation is less than 0.001, as observed in the data
presented, this is evidence of a strong relationship between the variables in a significant
way. The p-value complements the Pearson correlation coefficient, providing information
on the reliability of the observed relationship.
A low p-value, together with a considerable correlation coefficient, supports the claim
that the use of generative artificial intelligence tools in teaching programming has a positive
impact on aspects such as concept understanding, motivation, and student interest [63,64].
Pearson’s r value for the relationship between question Q2 (perceived ease of un-
derstanding complex topics) and question Q1 (level of understanding of basic concepts)
is 0.598. This determines a moderate positive correlation; that is, students who report a
better understanding of basic concepts also tend to find it easier to understand complex
topics. This is further evidence that the use of GenAI tools such as Gemini in Google
Colab can facilitate the understanding of basic concepts and the transition to the cognitive
development of more advanced concepts.
Pearson’s r coefficient between question Q3 (interest in learning programming) and
question Q1 (level of understanding of basic concepts) is 0.579, which also indicates a
moderate positive correlation. This is evidence that those students who feel that they have
improved their understanding tend to show a greater interest in programming. This further
implies that improving understanding can have a positive effect on motivation and interest
in the subject.
The relationship between question Q4 (motivation when using GenAI tools) and ques-
tion Q1 (level of understanding of basic concepts) shows a coefficient of 0.616, indicating a
moderate-high positive correlation. Students who feel more motivated to learn program-
ming are those who also report a better understanding of the concepts. This is evidence
that motivation and understanding are intimately connected, which is a direct result of
using GenAI like Gemini in Google Colab in learning [13,57].
The analysis of the relationship between question Q5 (satisfaction with fulfillment of
expectations) and question Q1 (level of understanding of basic concepts) shows a Pearson’s r
value of 0.532. Although this value is lower than in other cases, it still indicates a significant
positive correlation. This is evidence that students who feel that their learning expectations
have been met are those who also have a better understanding of basic concepts. This
reflects that the effectiveness of the GenAI Gemini with Google Colab in the classroom is
aligned with students’ expectations.
The relationship between question Q6 (perception of support received) and question
Q1 (level of understanding of basic concepts) has a coefficient of 0.602. This value indicates
a significant positive correlation, suggesting that students who feel that the GenAI Gemini
with Google Colab has exceeded their expectations in terms of support also tend to have a
better understanding of programming concepts. This reinforces the idea that the imple-
mentation of GenAI tools meets expectations, and offers additional support in learning.
The use of Spearman’s rho is a non-parametric measure that assesses the correlation
between two variables, considering their ordinal relationship. In the analysis conducted,
the results showed significant Spearman’s rho values, which ranged from 0.574 to 0.782,
indicating moderate to strong positive correlations between the variables evaluated. These
results show that as students experience a greater understanding of programming con-
cepts, and an increase in their motivation, they also tend to report greater interest and
Educ. Sci. 2024, 14, 1330 15 of 19

satisfaction with the use of generative artificial intelligence tools. The associated p-values,
all less than 0.001, reinforce the statistical significance of these correlations, implying that
it is highly unlikely that these observed relationships are due to chance. Taken together,
these findings highlight the positive impact of generative artificial intelligence on program-
ming instruction, supporting the idea that its integration can facilitate more effective and
motivating learning.
In the results obtained, Kendall’s Tau B values ranged from 0.540 to 0.748, evidencing
a moderate to strong positive correlation between the variables assessed, similar to that
observed with Spearman’s rho. The corresponding p-values, all less than 0.001, indicate
that these correlations are statistically significant, reinforcing confidence that the findings
reflect real relationships, and are not the product of chance. This implies that, as with
the other correlation measures, there is a clear association between the use of generative
artificial intelligence tools and improvements in students’ understanding, motivation and
interest in the subject, emphasizing the value of GenAI in education.
The use of GenAI tools in teaching programming is linked to significant improve-
ments in student understanding, motivation, and satisfaction. In addition to facilitating
the assimilation of concepts, it also increases interest and motivation, meeting and even
exceeding student expectations. This reality is of great importance when integrating GenAI
Gemini with Google Colab in the teaching of programming, thus promoting more effective
and stimulating learning.

5. Discussion
The findings of this study highlight the significant impact of generative artificial intel-
ligence tools in teaching programming. Through an exhaustive analysis, it became evident
that the use of GenAI not only improves the comprehension of fundamental concepts, but
also boosts students’ motivation and interest [37]. The correlations observed, both Pearson’s
r and Spearman’s rho and Kendall’s Tau B, indicate that as students interact with technolo-
gies such as Google Colab and Gemini, they experience greater satisfaction and a higher
fulfillment of expectations in their learning process [38]. This positive effect suggests that
GenAI can play an important and significant role in transforming programming education
by offering a more dynamic and accessible approach to tackling complex topics [27].
However, it is essential to consider the implications of these results for curriculum
design and professor training. The integration of GenAI in the classroom should not be
merely mechanical or technical; it is critical that educators understand how to use these
tools effectively to maximize their educational potential. In addition, more research should
be conducted to explore how different contexts and student populations may influence
the effectiveness of these tools [26]. As technology advances, it is critical to adapt teaching
strategies to ensure that all students benefit from innovations in learning, thereby fostering
a more inclusive, relevant, and meaningful education.
Indeed, this study corroborates the works [6–8], revealing that generative artificial
intelligence tools, such as Gemini at Google Colab, have a significant effect on the under-
standing of fundamental concepts. Students report an improvement in their assimilation
of basic topics, suggesting that GenAI facilitates the teaching of essential fundamentals
such as vectors and arrays, as well as the design of control structures, selection, and
repetition [34,37]. This understanding is fundamental for students, as academic success
depends on it, and serves as a solid foundation for tackling more advanced concepts, thus
highlighting the importance of integrating these tools into the programming curriculum
while avoiding unethical patterns of behavior [11].
In terms of ease of learning complex topics, GenAI proves to be a valuable resource.
The results indicate that students find learning difficult content more accessible compared
to traditional methods. This finding suggests that GenAI acts as a support and transforms
the way students interact with the material, promoting more dynamic and comprehensible
professor-supervised learning. In addition, the integration of GenAI in the classroom
Educ. Sci. 2024, 14, 1330 16 of 19

seems to spark a greater interest in programming, which is a key motivating factor in the
educational process [59].
Finally, when examining learning expectations, most students feel that the GenAI not
only meets, but often exceeds, their expectations. This positive perception extends to the
support they perceive they receive from these tools in their learning process. Satisfaction
with the use of GenAI in teaching programming demonstrates the effectiveness of these
technologies in improving both student motivation and engagement which, in turn, has a
lasting impact on their academic and professional development in programming [23].

6. Conclusions
The results of this study conclusively show that the incorporation of generative arti-
ficial intelligence tools in the teaching of programming has a significant positive impact
on student learning, motivation, and interest. The observed correlations suggest that the
use of technologies such as Gemini in Google Colab not only facilitates the understanding
of fundamental concepts, but also fosters greater satisfaction in the learning process. This
finding highlights the importance of integrating GenAI as an essential component in pro-
gramming curricula, which can transform the educational experience by making it more
engaging and effective.
However, it is essential to recognize that the successful implementation of these
tools requires adequate training for educators, ethical training for students and users,
an open attitude towards their possibilities, and a curriculum design capable of taking
full advantage of GenAI capabilities. By continuing to research and adjust innovative
methodologies in course content, it is ensured that programming education not only
remains relevant in an ever-evolving technological world, but also prepares students
effectively to meet the challenges of the future.
The use of these tools has been found to significantly improve students’ understanding
of fundamental programming concepts. This improvement translates into greater assimila-
tion of key content which, in turn, establishes a solid foundation for learning more complex
topics. Also, by analyzing how generative artificial intelligence facilitates the learning of
difficult topics, it has been shown that students feel better able to tackle advanced content
compared to traditional methods.
The integration of GenAI has also significantly influenced students’ interest and
motivation towards programming, which has been reflected in their greater willingness
to participate in related academic activities. Finally, it has been evidenced that students’
learning expectations are not only met, but in many cases exceeded, highlighting the added
value that GenAI brings to the educational experience. Taken together, these findings
emphasize the importance of incorporating advanced technologies into programming
instruction to improve student engagement and satisfaction.
In conclusion, this work has made it possible to evaluate the impact of generative
artificial intelligence in the teaching of programming, revealing that tools such as Gemini in
Google Colab have positive effects on various aspects of learning. The authors’ intention, in
addition to adding to the existing literature on the field of generative AI in education, is to
open spaces for meaningful learning activities that enhance the classroom experience with
the use of novel tools. In addition, future research focuses on exploring best practices for
integrating these technologies ethically, and how they can be adapted to diverse populations
of students with different professional profiles.

Author Contributions: Conceptualization, J.L.-I., J.M.-R. and R.A.-C.; methodology, J.L.-I. and R.A.-C.;
software, C.A.-M.; validation, J.L.-I., J.M.-R. and R.A.-C.; formal analysis, C.A.-M.; investigation, J.L.-I.
and R.A.-C.; resources, J.M.-R. and R.A.-C.; data curation, J.L.-I.; writing—original draft preparation,
C.A.-M.; writing—review and editing, J.L.-I. and R.A.-C.; visualization, J.M.-R.; supervision, J.L.-I.;
project administration, C.A.-M.; funding acquisition, J.L.-I. All authors have read and agreed to the
published version of the manuscript.
Funding: This research received no external funding.
Educ. Sci. 2024, 14, 1330 17 of 19

Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: The data presented in this study are available on request from the
corresponding author.
Acknowledgments: The authors would like to thank the authorities of the Universidad Politécnica
Salesiana in the city of Guayaquil, Ecuador, for their support in the project “Design of a research
training methodology for new university professors and administrative staff” (with the acronym
FINVE+P) of the GIEACI group, with Resolution No. 032-002-2024-02-27.
Conflicts of Interest: The authors declare no conflicts of interest.

References
1. González-Calatayud, V.; Prendes-Espinosa, P.; Roig-Vila, R. Artificial Intelligence for Student Assessment: A Systematic Review.
Appl. Sci. 2021, 11, 5467. [CrossRef]
2. Hazarika, H.; Achumi, A.S.; Konch, P.K. Exploring the Future Possibility of Generative AI in Libraries: A Comprehensive Study;
INFLIBNET Centre: Gujarat, India, 2024.
3. Cabero-Almenara, J.; Palacios-Rodríguez, A.; Loaiza-Aguirre, M.I.; Rivas-Manzano, M.D. Acceptance of Educational Artificial
Intelligence by Teachers and Its Relationship with Some Variables and Pedagogical Beliefs. Educ. Sci. 2024, 14, 740. [CrossRef]
4. Hwang, G.J.; Xie, H.; Wah, B.W.; Gašević, D. Vision, challenges, roles and research issues of Artificial Intelligence in Education.
Comput. Educ. Artif. Intell. 2020, 1, 100001. [CrossRef]
5. Kalota, F. A Primer on Generative Artificial Intelligence. Educ. Sci. 2024, 14, 172. [CrossRef]
6. Rane, N.; Choudhary, S.; Rane, J. Gemini versus ChatGPT: Applications, performance, architecture, capabilities, and implementa-
tion. J. Appl. Artif. Intell. 2024, 5, 69–93. [CrossRef]
7. Imran, M.; Almusharraf, N. Google Gemini as a next generation AI educational tool: A review of emerging educational
technology. Smart Learn. Environ. 2024, 11, 22. [CrossRef]
8. Edwards, K.; Scalisi, C.; DeMars-Smith, J.; Lee, K. Google Colab for Teaching CS and ML. In Proceedings of the 55th ACM
Technical Symposium on Computer Science Education V. 2, New York, NY, USA, 15–18 March 2023; SIGCSE: Pittsburgh, PA,
USA, 2024; p. 1925. [CrossRef]
9. Salas-Pilco, S.Z.; Yang, Y. Artificial intelligence applications in Latin American higher education: A systematic review. Int. J.
Educ. Technol. High. Educ. 2022, 19, 21. [CrossRef]
10. Ayala-Carabajo, R.; Llerena-Izquierdo, J. Bibliometric Review on a Hybrid Learning Model with VLEs from a Higher Education
Context. In Systems, Smart Technologies and Innovation for Society; Salgado-Guerrero, J.P., Vega-Carrillo, H.R., García-Fernández, G.,
Robles-Bykbaev, V., Eds.; Springer Nature: Cham, Switzerland, 2024; pp. 148–157. [CrossRef]
11. Al Lawati, B.H.; Lee, J.; Xu, L.Z.; Ow, T.T. Embracing GenAI In The Classroom: Getting Business Students Excited About
Information Technology. AIS TREO Papers. 2024. Available online: https://aisel.aisnet.org/treos_amcis2024/144/ (accessed on
15 January 2020).
12. Jayakody, R.; Dias, G. Performance of Recent Large Language Models for a Low-Resourced Language. In Proceedings of the 2024
International Conference on Asian Language Processing (IALP), Hohhot, China, 4–6 August 2024; pp. 162–167. [CrossRef]
13. Islam, R.; Ahmed, I. Gemini-the most powerful LLM: Myth or Truth. In Proceedings of the 2024 5th Information Communication
Technologies Conference (ICTC), Nanjing, China, 10–12 May 2024; pp. 303–308. [CrossRef]
14. Perera, P. Preparing to Revolutionize Education with the Multi-Model GenAI Tool Google Gemini? A Journey towards Effective
Policy Making. J. Adv. Educ. Philos. 2024, 7, 246–253. [CrossRef]
15. Verdugo-Castro, S.; García-Holgado, A.; Sánchez-Gómez, M.C. The gender gap in higher STEM studies: A systematic literature
review. Heliyon 2022, 8, e10300. [CrossRef]
16. Lin, H. Influences of Artificial Intelligence in Education on Teaching Effectiveness: The Mediating Effect of Teachers’ Perceptions
of Educational Technology. Int. J. Emerg. Technol. Learn. (iJET) 2022, 17, 144–156. [CrossRef]
17. Team, G.; Georgiev, P.; Lei, V.I.; Burnell, R.; Bai, L.; Gulati, A.; Tanzer, G.; Vincent, D.; Pan, Z.; Wang, S.; et al. Gemini 1.5:
Unlocking multimodal understanding across millions of tokens of context. arXiv 2024, arXiv:2403.05530.
18. Saeidnia, H.R. Welcome to the Gemini era: Google DeepMind and the information industry. Libr. Tech News 2023, ahead-of-print.
[CrossRef]
19. Fernández-Herrero, J. Evaluating Recent Advances in Affective Intelligent Tutoring Systems: A Scoping Review of Educational
Impacts and Future Prospects. Educ. Sci. 2024, 14, 839. [CrossRef]
20. Lye, C.Y.; Lim, L. Generative Artificial Intelligence in Tertiary Education: Assessment Redesign Principles and Considerations.
Educ. Sci. 2024, 14, 569. [CrossRef]
21. Hinojo-Lucena, F.J.; Aznar-Díaz, I.; Cáceres-Reche, M.P.; Romero-Rodríguez, J.M. Artificial Intelligence in Higher Education: A
Bibliometric Study on its Impact in the Scientific Literature. Educ. Sci. 2019, 9, 51. [CrossRef]
22. López-Chila, R.; Llerena-Izquierdo, J.; Sumba-Nacipucha, N.; Cueva-Estrada, J. Artificial Intelligence in Higher Education: An
Analysis of Existing Bibliometrics. Educ. Sci. 2024, 14, 47. [CrossRef]
Educ. Sci. 2024, 14, 1330 18 of 19

23. Hou, W.; Ji, Z. A systematic evaluation of large language models for generating programming code. arXiv 2024, arXiv:2403.00894.
24. Limna, P.; Kraiwanit, T. Google Gemini’s Influence on Workplace Dynamics in Bangkok. Hum. Behav. Dev. Soc. 2024, 25, 126–136.
[CrossRef]
25. Perezchica-Vega, J.E.; Sepúlveda-Rodríguez, J.A.; Román-Méndez, A.D. Inteligencia artificial generativa en la educación superior:
usos y opiniones de los profesores. Eur. Public Soc. Innov. Rev. 2024, 9, 1–20. [CrossRef]
26. Hernández González, M.; Ramos Quiroz, J.M.; Chávez Maciel, F.J.; Trejo Cázares, M.d.C. Ventajas y riesgos de la Inteligencia
Artificial Generativa desde la percepción de los estudiantes de educación superior en México. Eur. Public Soc. Innov. Rev. 2024,
9, 1–19. [CrossRef]
27. Padilla Piernas, J.M.; Martín-García, M.d.M. Impacto y Perspectivas de la Inteligencia Artificial Generativa en la Educación
Superior: Un Estudio sobre la Percepción y Adopción Docente usando el modelo AETGE/GATE. Eur. Public Soc. Innov. Rev. 2024,
9, 1–21. [CrossRef]
28. Solano Hilario, C.; Belinda Ccope Jaucha, K.; La Rosa de Benavides, L.M.A.; Paul Medina Perez, G. Generative Artificial
Intelligence and its impact on the educational quality of students: Systematic Review. In Proceedings of the LACCEI international
Multi-conference for Engineering, Education and Technology, San Jose, Costa Rica, 17–19 July 2024; pp. 1–11. [CrossRef]
29. García Peñalvo, F.J.; Llorens-Largo, F.; Vidal, J. La nueva realidad de la educación ante los avances de la inteligencia artificial
generativa. RIED-Rev. Iberoam. Educ. Distancia 2024, 27, 9–39. [CrossRef]
30. Schiavo, G.; Businaro, S.; Zancanaro, M. Comprehension, apprehension, and acceptance: Understanding the influence of literacy
and anxiety on acceptance of artificial Intelligence. Technol. Soc. 2024, 77, 102537. [CrossRef]
31. Solomovich, L.; Abraham, V. Exploring the influence of ChatGPT on tourism behavior using the technology acceptance model.
Tour. Rev. 2024, ahead-of-print. [CrossRef]
32. Henrique-Sanches, B.C.; Sabage, L.; Costa, R.R.d.O.; Almeida, R.G.d.S.; Moron, R.A.; Mazzo, A. Implications of practical activities
in the Skills and Simulation Laboratory on students’ motivation and feelings. Rev.-Lat.-Am. Enferm. 2023, 31. [CrossRef]
33. Vitale, A.; Dello Iacono, U. Usando robots sociales como tecnología educativa inclusiva para el aprendizaje de matemáticas a
través de la narración. Eur. Public Soc. Innov. Rev. 2024, 9, 1–17. [CrossRef]
34. Rossettini, G.; Rodeghiero, L.; Corradi, F.; Cook, C.; Pillastrini, P.; Turolla, A.; Castellini, G.; Chiappinotto, S.; Gianola, S.; Palese,
A. Comparative accuracy of ChatGPT-4, Microsoft Copilot and Google Gemini in the Italian entrance test for healthcare sciences
degrees: A cross-sectional study. BMC Med. Educ. 2024, 24, 694 [CrossRef]
35. Fabijan, A.; Zawadzka-Fabijan, A.; Fabijan, R.; Zakrzewski, K.; Nowosławska, E.; Kosińska, R.; Polis, B. Automated MRI Video
Analysis for Pediatric Neuro-Oncology: An Experimental Approach. Appl. Sci. 2024, 14, 8323. [CrossRef]
36. Schei, O.M.; Møgelvang, A.; Ludvigsen, K. Perceptions and Use of AI Chatbots among Students in Higher Education: A Scoping
Review of Empirical Studies. Educ. Sci. 2024, 14, 922. [CrossRef]
37. Alnasib, B.N.; Alharbi, N.S. Challenges and Motivation: Assessing Gemini‘s Impact on Undergraduate EFL Students in Classroom
Settings. World J. Engl. Lang. 2024, 14, 501– 514. [CrossRef]
38. Zichar, M.; Papp, I. Contribution of Artificial Intelligence (AI) to Code-Based 3D Modeling Tasks. Designs 2024, 8, 104. [CrossRef]
39. Torres-Peña, R.C.; Peña-González, D.; Chacuto-López, E.; Ariza, E.A.; Vergara, D. Updating Calculus Teaching with AI: A
Classroom Experience. Education Sciences 2024, 14, 1019. [CrossRef]
40. Yoseph, E.T.; Gonzalez-Suarez, A.D.; Lang, S.; Desai, A.; Hu, S.S.; Zygourakis, C.C. Patient perspectives on AI: A pilot study
comparing large language model and physician-generated responses to routine cervical spine surgery questions. Artif. Intell.
Surg. 2024, 4, 267–277. [CrossRef]
41. Almassaad, A.; Alajlan, H.; Alebaikan, R. Student Perceptions of Generative Artificial Intelligence: Investigating Utilization,
Benefits, and Challenges in Higher Education. Systems 2024, 12, 385. [CrossRef]
42. Hanci, V.; Ergün, B.; Gül, C.; Uzun, Ö.; Erdemir, I.; Hanci, F.B. Assessment of readability, reliability, and quality of ChatGPT® ,
BARD® , Gemini® , Copilot® , Perplexity® responses on palliative care. Medicine 2024, 103, e39305. [CrossRef]
43. Karaca, M.F. Is Artificial Intelligence able to Produce Content Appropriate for Education Level? A Review on ChatGPT and
Gemini. In Proceedings of the Cognitive Models and Artificial Intelligence Conference, Istanbul, Turkiye, 25–26 May 2024;
pp. 208–213. [CrossRef]
44. Gomez-Cabello, C.A.; Borna, S.; Pressman, S.M.; Haider, S.A.; Forte, A.J. Large Language Models for Intraoperative Decision
Support in Plastic Surgery: A Comparison between ChatGPT-4 and Gemini. Medicina 2024, 60, 957. [CrossRef]
45. Kharchenko, Y.V.; Babenko, O.M. Advantages and limitations of large language models in chemistry education: A comparative
analysis of ChatGPT, Gemini and Copilot. In Proceedings of the Free Open-Access Proceedings for Computer Science Workshops,
Lviv, Ukraine, 23 September 2024; Volume 3781, pp. 42–59.
46. Sonmezoglu, B.G.; Sonmezoglu, H.I. Comparative Analysis of AI Chatbots Chat GPT, Gemini, and Copilot’s Answers to Common
Cataract Questions. Pak. J. Ophthalmol. 2024, 40, 370–375. [CrossRef]
47. Meyer, A.; Soleman, A.; Riese, J.; Streichert, T. Comparison of ChatGPT, Gemini, and le Chat with physician interpretations of
medical laboratory questions from an online health forum. Clin. Chem. Lab. Med. 2024, 62, 2425–2434. [CrossRef]
48. Durmaz Engin, C.; Karatas, E.; Ozturk, T. Exploring the Role of ChatGPT-4, BingAI, and Gemini as Virtual Consultants to Educate
Families about Retinopathy of Prematurity. Children 2024, 11, 750. [CrossRef]
49. Farghal, M.; Haider, A.S. Translating classical Arabic verse: Human translation vs. AI large language models (Gemini and
ChatGPT). Cogent Soc. Sci. 2024, 10, 2410998. [CrossRef]
Educ. Sci. 2024, 14, 1330 19 of 19

50. Is, E.E.; Menekseoglu, A.K. Comparative performance of artificial intelligence models in rheumatology board-level questions:
Evaluating Google Gemini and ChatGPT-4o. Clin. Rheumatol. 2024, 43, 3507–3513. [CrossRef] [PubMed]
51. Lopez-Chila, R.; Mora-Saltos, N.; Cedeño-Tello, A.; Llerena-Izquierdo, J. A Learning Resource Management Model for high-
enrollment Programming courses in Engineering. In Proceedings of the 2023 International Conference on Electrical, Communica-
tion and Computer Engineering (ICECCE), Dubai, United Arab Emirates, 30–31 December 2023; pp. 1–6. [CrossRef]
52. Chan, W.K.; Yu, Y.T.; Keung, J.W.; Lee, V.C.S. Toward AI-assisted Exercise Creation for First Course in Programming through
Adversarial Examples of AI Models. In Proceedings of the 2023 IEEE 35th International Conference on Software Engineering
Education and Training (CSEE&T), Tokyo, Japan, 7–9 August 2023; pp. 132–136. [CrossRef]
53. Farah, J.C.; Spaenlehauer, B.; Ingram, S.; Purohit, A.K.; Holzer, A.; Gillet, D. Harnessing Rule-Based Chatbots to Support Teaching
Python Programming Best Practices BT - Towards a Hybrid, Flexible and Socially Engaged Higher Education. In Towards a Hybrid,
Flexible and Socially Engaged Higher Education; Auer, M.E., Cukierman, U.R., Vendrell Vidal, E., Tovar Caro, E., Eds.; Springer:
Cham, Switzerland, 2024; pp. 455–466.
54. Cedeño-Tello, A.; Llerena-Izquierdo, J. Homogeneity of Engineering Courses from an Assignment Management Model in Virtual
Learning Environments. In International Conference on Science, Technology and Innovation for Society; Springer: Berlin/Heidelberg,
Germany, 2023; pp. 439–447. [CrossRef]
55. Chiang, Y.H.V.; Lin, Y.C.; Chen, N.S. Developing a Course-Specific Chatbot Powered by Generative AI for Assisting Students’
Learning in a Programming Course. In Proceedings of the 2024 IEEE International Conference on Advanced Learning Technologies
(ICALT), Nicosia, Cyprus, 1–4 July 2024; pp. 182–184. [CrossRef]
56. Llerena-Izquierdo, J. Virtual Classroom Design Model and Its Relation to Student Motivation and Performance in a Moodle
Learning Environment During the Emergency of COVID-19. In Doctoral Symposium on Information and Communication Technologies—
DSICT; Lecture Notes in Electrical Engineering; Berrezueta, S., Abad, K. Eds.; Springer: Cham, Switzerland, 2022; Volume 846,
pp. 21–32. [CrossRef]
57. Shan, R.; Ming, Q.; Hong, G.; Wu, H. Benchmarking the hallucination tendency of google gemini and moonshot kimi. OSF-
Preprints 2024. [CrossRef]
58. Studio, A.; Vertex, A. Developments in AI, Programming, Web, Security, Virtual and Augmented Reality, and Quantum
Computing. JSTC 2024, 35, 74–76. [CrossRef]
59. Yang, Q. Systematic Evaluation of AI-Generated Python Code: A Comparative Study Across Progressive Programming Tasks; Research
Square Company: Durham, NC, USA, 2024.
60. Llerena-Izquierdo, J. Adaptation of the curriculum in relation to student learning outcomes in initial programming courses. In
Proceedings of the 2023 IEEE World Engineering Education Conference (EDUNINE), Bogota, Colombia, 12–15 March 2023; IEEE:
Piscataway, NJ, USA; pp. 1–6. [CrossRef]
61. Yu, C. Unlocking the Full Potential of AI Chatbots: A Guide to Maximizing Your Digital Companions. In Proceedings of the
Companion Proceedings of the 32nd ACM International Conference on the Foundations of Software Engineering, New York, NY,
USA, 15–19 July 2024; FSE: Hong Kong, 2024; pp. 680–682. [CrossRef]
62. Deriba, F.; Sanusi, I.T.; O.Campbell, O.; Oyelere, S.S. Computer Programming Education in the Age of Generative AI: Insights
from Empirical Research. SSRN Electron. J. 2024, 1–11. [CrossRef]
63. Sağın, F.G.; Özkaya, A.B.; Tengiz, F.; Geyik, Ö.G.; Geyik, C. Current evaluation and recommendations for the use of artificial
intelligence tools in education. Degruyter 2023, 48, 620–625. [CrossRef]
64. Herden, O. Integration of Chatbots for Generating Code into Introductory Programming Courses. 2024; pp. 6–10. Available
online: https://conference.pixel-online.net/library_scheda.php?id_abs=6673 (accessed on 15 January 2020).

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.

View publication stats