The Impact of AI Tools on Software Development: A Case Study with

GitHub Copilot and Other AI Assistants

Sergio Cavalcante

, Erick Ribeiro

and Ana Carolina Oran

Fundac¸

ao Paulo Feitoza – FPFtech, Brazil

Institute of Computing, Universidade Federal do Amazonas - UFAM, Brazil

Keywords:

ChatGPT, GitHub Copilot, Code Generation, Empirical Study, Software Engineering, Software Development,

Artiﬁcial Intelligence.

Abstract:

Background - With the increasing complexity of software projects and the demand for rapid and high-quality

deliveries, Generative Artiﬁcial Intelligence (GenAI) tools have emerged as powerful allies in software devel-

opment. Objective - This study aims to evaluate the impact of using Code Generation Assistants—such as

GitHub Copilot, ChatGPT, and Gemini—in software development environments. Method - We conducted a

satisfaction survey with 57 volunteers in an R&D organization, including developers, test analysts, and prod-

uct owners, collecting quantitative and qualitative data on the use of these tools. Results - The results indicate

that the use of these tools signiﬁcantly increased productivity, improved code quality, and accelerated profes-

sional learning. Additionally, it facilitated the automation of repetitive tasks, allowing focus on more complex

challenges. However, challenges such as the need for constant review of generated code and the risk of exces-

sive dependency were identiﬁed. Conclusion - We conclude that, despite the challenges, GenAI tools have a

signiﬁcant positive impact on software development, and organizational support is crucial to maximize their

beneﬁts.

1 INTRODUCTION

With the increasing complexity of software projects

and the demand for rapid and high-quality deliver-

ies, Generative Artiﬁcial Intelligence (GenAI) tools

have emerged as powerful allies in the ﬁeld of soft-

ware development (Ziegler et al., 2022). Developers

and software engineers are always seeking best prac-

tices, tools, and ways to improve work processes (Ra-

jlich, 2014). Recently, a notable trend has been the

adoption of Code Generation Assistants. Among the

code assistants available in the market, GitHub Copi-

lot

, ChatGPT

, and Gemini

stand out, all based on

GenAI, with the potential to transform the way devel-

opment professionals code software.

Currently, the use of systems based on GenAI is

already being explored in other areas outside the cor-

porate environment and in the context of software de-

velopment. For example, in the ﬁeld of education

through the automated generation of teaching materi-

als (Lim et al., 2023), in the creation of personalized

https://github.com/features/copilot/

https://chat.openai.com/

https://gemini.google.com/

tutorials (Moon et al., 2023), and in assisting code

generation for computer science students (Lira et al.,

2024). The growing adoption and massive use of this

generative technology are due to its ability to gener-

ate high-quality content, similar to what a human be-

ing would generate. As a result, the potential applica-

tions of GenAI are expanding rapidly, and its impact

on various industries could be transformative, paving

the way for more efﬁcient workﬂows and innovative

solutions across multiple domains.

In the corporate context, studies on the collabo-

rative partnership between software engineers and in-

telligent code generation assistants and AI tools are

still incipient. There is a need for works that ex-

plore the possible advantages and disadvantages ob-

tained with the use of this technology, as well as the

main changes for the software development industry.

In this scenario of rapid changes provided by GenAI,

this work aims to answer the following central ques-

tion: How are software engineers being impacted

by the use of intelligent code generation assistants

and AI tools?

The research presented in this article aims to eval-

uate the impact of using Code Generation Assistants

Cavalcante, S., Ribeiro, E. and Oran, A. C.

The Impact of AI Tools on Software Development: A Case Study with GitHub Copilot and Other AI Assistants.

DOI: 10.5220/0013294700003929

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

In Proceedings of the 27th International Conference on Enterprise Information Systems (ICEIS 2025) - Volume 2, pages 245-252

ISBN: 978-989-758-749-8; ISSN: 2184-4992

245

and AI tools in software development environments,

through the application of a satisfaction survey with

57 volunteers. The relevance of this study lies in

the growing adoption of these tools, which promise to

streamline the coding process, reduce errors, and in-

crease productivity. By investigating developers’ per-

ceptions regarding the beneﬁts, limitations, and chal-

lenges of using these assistants, we can offer valu-

able insights into the evolution of these technologies.

This work seeks not only to measure the direct impact

on professionals’ efﬁciency but also to contribute to a

deeper understanding of the acceptance and expecta-

tions around these tools in the daily life of software

engineering.

This article is organized as follows: Section 2

presents works that explore the use of the main cur-

rent code generation tools. In Section 3, the method-

ology applied in this study is presented. Sections 4

and 5 present the results obtained, including insights

and observations. Finally, in Section 6, the ﬁnal con-

siderations and future works are described.

2 RELATED WORKS

This section addresses works that have analyzed the

use of automatic code generation tools in the ﬁeld of

software development from different perspectives.

In (Imai, 2022), a pair programming experi-

ment was conducted involving 21 participants, using

GitHub Copilot versus a human partner. The study

measured productivity based on lines of code pro-

duced and code quality by the lines removed. The

researchers concluded that Copilot helps generate a

higher number of lines of code but, in return, results

in a higher number of lines deleted. Although it is a

study focused on the amount of code, it is important

to note that the number of lines of code is not nec-

essarily indicative of code quality; our work did not

follow this line of evaluation.

In (Peng et al., 2023), an experiment was con-

ducted by Microsoft researchers involving 95 pro-

grammers. The participants were tasked with creating

an HTTP server in the JavaScript language, with the

option to seek help online for the challenges faced.

The programmers were divided into two groups: one

with access to GitHub Copilot (the treatment group)

and one without (the control group). The results

showed that the treatment group completed their tasks

faster, even with less experienced developers. Our ex-

periment also includes developers but is not limited to

the perspective of technical people working to solve a

problem.

In (Yetis¸tiren et al., 2023), GitHub Copilot was

compared with AWS CodeWhisperer and ChatGPT

in various code quality perspectives, such as validity,

correctness, security, reliability, and maintainability.

They found that the latest versions of these tools in-

ﬂuence the ability to generate correct code. Among

the ﬁndings, ChatGPT was the most successful tool,

while Amazon CodeWhisperer was the worst. In the

comparison between GitHub Copilot and CodeWhis-

perer, Copilot performed better. In light of this work,

in our study, we also chose to evaluate multiple tools,

discarding CodeWhisperer.

3 METHODOLOGY

This study aimed to explore the impact of AI tools

on software development within an R&D organiza-

tion focused on delivering tailored software solutions

to clients across various industries. The methodology

included deﬁning research objectives, selecting par-

ticipants based on their use of AI tools, and creating

a survey focused on areas such as usage frequency,

perceived beneﬁts, and challenges.

3.1 Study Planning

The research was designed to capture both quantita-

tive and qualitative data, aiming to provide a com-

prehensive view of the impacts perceived by users of

these tools. The methodology was divided into three

main phases:

1. Deﬁnition of Research Objectives. The goal

was to explore how code generation tools, such as

GitHub Copilot, ChatGPT, and Gemini, are being

used in the organization and what is the perceived

impact by developers, test analysts, product own-

ers (POs), and other roles.

2. Sample Selection. The survey was sent to profes-

sionals from different areas, including developers

(junior, mid-level, and senior), test analysts, and

product owners. Participants were selected based

on their use of the mentioned tools, either through

organizational licenses or personal use. A total of

57 valid responses were collected.

3. Questionnaire Design. The questionnaire was

developed based on questions raised in the refer-

ence study

and adapted to the reality of the orga-

nization. The ChatGPT tool was used to assist in

creating the questions, ensuring they were com-

prehensive and focused on speciﬁc points such as

https://www.tabnine.com/blog/github-copilot-for-

business/

ICEIS 2025 - 27th International Conference on Enterprise Information Systems

246

productivity, code quality, job satisfaction, learn-

ing, and challenges in using the tools. The ques-

tionnaire was divided into ﬁve main sections:

(a) Frequency and Type of Use. Questions to

measure the frequency of use and whether pro-

fessionals were using licenses provided by the

company, free versions, or purchased with per-

sonal resources.

(b) Tools Used Identify which other AI tools were

being used besides GitHub Copilot.

perceived beneﬁts in using the tools, such as

increased productivity, code quality, and learn-

ing.

(d) Overall Impact on Work. Assessment of how

the tools inﬂuenced meeting deadlines, focus

on problem-solving, and impact on overall sat-

isfaction.

(e) Challenges and Disadvantages. Open-ended

questions to identify the main challenges and

disadvantages perceived in using AI tools.

3.2 Conducting the Study

The study was conducted over three months by three

researchers—two from the organization and one from

academia—and involved the application of a ques-

tionnaire distributed through Microsoft Forms, a tool

commonly used by the organization for feedback

collection. The questionnaire targeted development

teams, QA, and product management, with empha-

sis on developers and roles such as product own-

ers (POs), who frequently use tools like ChatGPT.

The sample consisted of 57 valid responses, with

participants reporting their frequency of AI tool us-

age, licensing types, perceived beneﬁts, and chal-

lenges. Quantitative data were collected using struc-

tured questions, including Likert scales, while qual-

itative data were obtained through open-ended ques-

tions, later categorized into themes such as ”tool de-

pendence” and ”quality of code suggestions.”

3.3 Data Analysis

The data were analyzed through a combination of

quantitative and qualitative analysis. The analysis

process followed these steps:

1. Quantitative Analysis. For the quantitative data,

Microsoft Excel was used to calculate frequencies

and distributions of responses. The analysis fo-

cused on identifying usage patterns, such as the

frequency with which the tools were used and the

most common type of license among participants.

Graphs were generated to illustrate the distribu-

tions of responses, such as the perceived impact

on productivity and code quality.

2. Qualitative Analysis. Open-ended responses

were analyzed using thematic categorization tech-

niques. Responses were grouped into categories

such as “productivity beneﬁts”, “increase in code

quality”, “technical challenges”, and “perceived

disadvantages”. This process allowed us to iden-

tify patterns in users’ perceptions and brought im-

portant insights for discussing the results.

3. Data Integration. Finally, qualitative and quanti-

tative data were integrated to provide a complete

view of the impact of AI tools in the work environ-

ment. Emphasis was given to cases where qualita-

tive responses corroborated quantitative ﬁndings,

such as reports on increased productivity and tool

dependence.

4 RESULTS

In this section, we present the ﬁndings of our study,

which aimed to evaluate the impact of using Code

Generation Assistants in software development envi-

ronments, through a satisfaction survey with 57 vol-

unteers. We analyzed both quantitative and qualitative

data to offer a comprehensive view of developers’ per-

ceptions regarding the beneﬁts, limitations, and chal-

lenges in using these tools.

4.1 Quantitative Analysis

The volunteers reported signiﬁcant improvements in

efﬁciency and productivity when using AI tools used

in the study.

4.1.1 Frequency of Use of the Tools

Among the 57 respondents, the majority reported fre-

quent use of AI tools (such as GitHub Copilot, Copi-

lot 365, ChatGPT, etc.), with emphasis on daily use.

The distribution of use was as in Figure 1.

The pie chart shows the frequency of AI tool us-

age among the 57 respondents, reﬂecting a high in-

tegration of these tools into their workﬂows. This

data highlights the frequent adoption of AI tools, em-

phasizing their role in improving productivity and

streamlining tasks in daily work routines.

4.1.2 Perceived Beneﬁts

When asked about the beneﬁts of using AI, most re-

sponses were positive, as shown in Figure 2, with

The Impact of AI Tools on Software Development: A Case Study with GitHub Copilot and Other AI Assistants

247

37%

47%

Daily Rarely

Several times a week Never

Once a week

Figure 1: Frequency of Use of the Tools.

A B

D E

Number of Respondents

A: Increased productivity

B: Reduction of time on repetitive tasks

C: Faster learning and knowledge acquisition

D: Increased creativity and problem-solving

E: Improvement in code quality

Figure 2: Perceived Beneﬁts Reported by Respondents.

participants citing improved efﬁciency and decision-

making.

The bar chart presents the perceived beneﬁts of

using AI tools in software development, highlight-

ing that the most signiﬁcant impact is on faster learn-

ing and knowledge acquisition (C), with 49 respon-

dents reporting this beneﬁt. This represents a large

portion of the respondents, indicating that AI tools

are strongly facilitating skill development. Increased

productivity (A) was the second most cited beneﬁt,

with 43 respondents, reﬂecting the tools’ contribu-

tion to efﬁciency in the workplace. Other notable

impacts include the reduction of time on repetitive

tasks (B) with 36 respondents and increased creativ-

ity and problem-solving (D) reported by 29 respon-

dents, which suggests that the tools not only save time

but also help developers focus on more complex and

innovative tasks. Improvement in code quality (E)

was mentioned by 25 respondents, indicating a pos-

itive, although slightly lesser, inﬂuence on the qual-

ity of work. These ﬁndings underline that AI tools

are highly valued across multiple dimensions, partic-

ularly for enhancing learning, productivity, and cre-

ativity.

4.1.3 Impact on Overall Work Experience

When asked about the overall impact of AI tools on

work.

The pie chart, shown in Figure 3, illustrates the

respondents’ perception of the overall impact of AI

tools on their work experience. This data suggests

that a vast majority of the respondents—48 out of 57

(84%) in total—have experienced some level of im-

provement in their daily work routines due to these

tools, reinforcing the idea that AI is contributing

meaningfully to enhancing productivity and job sat-

isfaction in software development environments.

44%

40%

16%

Improved signiﬁcantly

Improved slightly

Had no impact

Figure 3: Impact Reported by Respondents.

4.1.4 Ability to Meet Deadlines

In terms of meeting deadlines, the distribution was as

in Figure 4.

The pie chart displays respondents’ perceptions

of how AI tools impacted their ability to meet dead-

lines. These results suggest that a majority of respon-

dents—43 out of 57 (75%)—recognized some level of

improvement in managing deadlines, likely due to the

efﬁciency and time-saving features of AI tools. The

signiﬁcant portion that acknowledged improvement

reﬂects the general sentiment that such tools help pro-

fessionals avoid delays by automating routine tasks

and providing quicker access to solutions, allowing

more focus on meeting deadlines.

ICEIS 2025 - 27th International Conference on Enterprise Information Systems

248

24%

51%

26%

Improved signiﬁcantly

Improved slightly

Had no impact

Figure 4: Ability to meet Deadlines.

4.1.5 Focus on Problem Solving

Asked if the tools helped focus on more complex

tasks, the distribution was as in Figure 5.f

18%

45%

32%

Strongly disagree Disagree

Neutral Agree

Strongly agree

Figure 5: Focus on Problem Solving.

The pie chart illustrates how respondents felt AI

tools helped them focus on problem-solving tasks.

The overall trend suggests that the majority—43 out

of 57 respondents (77%)—felt that the tools helped

them focus on the core of their work, reducing time

spent on tedious tasks. By allowing engineers to

bypass less engaging or repetitive aspects of devel-

opment, these tools contributed to a more satisfy-

ing work experience, as developers could concentrate

more on problem-solving and innovation, potentially

improving overall job satisfaction.

4.2 Qualitative Analysis

To deepen the understanding of professionals’ experi-

ences with AI tools, we conducted a qualitative anal-

ysis of open-ended responses. The feedback was cat-

egorized into positive aspects and challenges faced,

highlighting key themes, which we discuss below.

Please note that only a subset of responses will be pre-

sented here. To access the complete set of responses,

please refer to the following link on GitHub

4.2.1 Positive Points

Code Generation and Code Search. Many respon-

dents cited increased productivity and development

speed as one of the main beneﬁts:

“Speciﬁc searches for the use of methods and

classes, almost automatic code documentation.

Sometimes it suggests more than I would like, but

those were exceptions.” – P18 (Data Scientist)

“It greatly accelerated my process when writing

repetitive codes and solving them more efﬁciently

instead of creating a function; it suggested a native

function that I did not know.” – P42 (Developer)

These comments highlight how AI tools help de-

velopers quickly ﬁnd code snippets and solve simple

doubts, saving valuable time.

Code Quality. Improvement in code quality was

also highlighted, citing better understanding, suc-

cinct, readable, and more efﬁcient code:

“Improved code quality because I can visualize other

possibilities. Improved documentation develop-

ment speed and also enabled testing new libraries

more agilely.” – P31 (Developer)

“Improvement in code quality following best prac-

tices. Creating, commenting, and compacting lad-

der programming.” – P32 (Automation Engi-

neer)

These responses suggest that AI tools not only

help code faster but also write better and more efﬁ-

cient code.

Learning. AI was perceived as a vector for learning:

“It mainly helped in solving problems involving

stacks that are generally not used in the lab; cur-

rently, I work on a project involving WPF and C#,

and I had never had contact with these technolo-

gies; the role of AI in increasing my learning pro-

cess in these stacks has been fundamental for my

delivery deadlines and conventional quality of the

codes I produce.” – P46 (Developer)

https://github.com/erickrribeiro/iceis-

2025/blob/main/answers.csv

The Impact of AI Tools on Software Development: A Case Study with GitHub Copilot and Other AI Assistants

249

These responses indicate that AI tools contribute

to faster learning and understanding of new technolo-

gies.

Repetitive Tasks. Faster and more focus on the main

problem:

“When working with backend and frontend, you of-

ten need to insert repetitive code. An AI tool iden-

tiﬁes such recurrence and already inserts the code,

increasing productivity.” – P26 (Developer).

Infra/DevOps. AI tools assist in creating scripts

and complex commands, streamlining infrastructure

tasks:

“Many things like scripts to create something in Ter-

raform, GitLab CI, infra tools in general context,

to make a sed command to change text inside an

Nginx conﬁguration ﬁle; it makes the regular ex-

pression, without me having to keep searching

how to get the value of a text to make the change

via regular expression; it already gives me how to

do it, and most of the time it’s correct” – P3 (De-

vOps)

QA. AI contributes to efﬁciency in creating test

scripts, speeding up the quality assurance process:

“Using Copilot in the development of my Scripts

greatly improves the speed of typing giant scripts,

as the functionality cites an example taking from

the context to speed up the process. Thus, greatly

helping in rapid development in the software test-

ing part” – P5 (QA Analyst)

Data Engineering. AI contributes to productivity in

creating Database scripts:

“In laborious tasks that would be done manually,

such as creating a schema in Spark with 300

columns” – P6 (Data Engineer)

Documentation/Proposals. Productivity and speed

in delivering proposals and documentation:

“For example, the use of AI tools in the preparation

of material made available to the Technical Re-

port team has a relevant impact, as it speeds up the

search for ﬁxed concepts and reference material,

as well as the possibility of improving the quality

of the prepared material” – P9 (People and Pro-

cess)

“Improved writing project proposals, email drafting,

etc.” – P51 (Project Manager)

These examples above show that the beneﬁts of AI

tools extend beyond coding, assisting in documenta-

tion and communication tasks.

Insights and Mentoring. Improvement in content

generation and ideas:

“Communication: I ask AI to apply Nonviolent

Communication in emails and others; Efﬁciency:

I talk to AI to get insights into problem-solving,

conﬂicts, and lessons learned ceremonies. Team

mentoring: review card writing and acceptance

criteria; Learning: when I want to apply some-

thing new with the team and need mentoring” –

P22 (People and Process)

“I don’t use AI just for automation but also for cre-

ation. In addition to improving and creating texts,

making lists, AI can generate ideas that evolve

into other ideas. This is extremely useful for solv-

ing various problems, from UX issues to tips on

how to deal with difﬁcult clients, as AI has the

ability to better understand contexts” – P29 (UX

Designer)

“It impacted practically all stages of a project’s ex-

ecution. I can focus more on the overall vision

of the project and delegate the creation of small

modules to AI.” – P25 (Product Owner)

These reﬂections above demonstrate how AI tools

can serve as a source of inspiration and guidance in

various aspects of work.

4.2.2 Negative Points

Need for Review and “Hallucinations”. Several re-

spondents expressed concerns about the accuracy of

the generated code:

“There would always be supervision due to common

hallucinations in free AI versions” – P8 (Devel-

oper)

“As a disadvantage, the approach suggested is not

always the most efﬁcient, but it serves as a good

guide” – P41 (Developer)

These comments above emphasize the importance

of reviewing the suggestions generated by AI tools to

ensure correctness.

Limited Context Understanding. Some users noted

that AI tools sometimes lack a complete understand-

ing of the project’s context:

“Regarding the disadvantages, it still leaves much to

be desired in recognizing the complete context of

the project’s codes as a whole. It still gets lost a

lot.” – P25 (Product Owner)

Generates Dependency. Some respondents men-

tioned the risk of excessive dependence on the tools:

“These LLMs are a tool. But people have become

dependent on them to the point of not wanting to

think anymore and going straight to consult them.

If they continue like this, they will stop being peo-

ple who think and become just ﬁreﬁghters.” – P30

(Developer)

ICEIS 2025 - 27th International Conference on Enterprise Information Systems

250

Thus, the data suggest that AI tools are playing a

signiﬁcant role in transforming software development

practices, with positive impacts on the efﬁciency and

satisfaction of the professionals involved.

5 DISCUSSION

The results of this study provide valuable insights into

the impact of using AI-based Code Generation Assis-

tants in software development. The initial motivation

was to understand how these tools are inﬂuencing the

work of software professionals, especially in terms of

productivity, code quality, and job satisfaction.

5.1 Perceived Beneﬁts

The quantitative data indicate that the adoption of AI

tools is signiﬁcant among the professionals surveyed,

with 84% of respondents using them daily or several

times a week. This high level of adoption suggests

that AI tools are becoming an integral part of devel-

opers’ workﬂows.

Most participants reported increased productiv-

ity (75%) and reduced time spent on repetitive tasks

(63%). These results are consistent with Related

Works section, which suggest that AI tools can sig-

niﬁcantly speed up the coding process and enhance

efﬁciency across various tasks. Additionally, 86% of

respondents experienced faster learning and knowl-

edge acquisition, highlighting the educational value

of these tools in supporting the continuous develop-

ment of professionals’ skills.

Improvement in code quality was highlighted by

44% of participants. Qualitative responses reinforce

this point, with developers like P31 (Developer) men-

tioning: “Improved code quality because I can visu-

alize other possibilities. . . and enabled testing new

libraries more agilely.” This suggests that AI tools not

only streamline code production but also encourage

more robust and efﬁcient coding practices.

Furthermore, the aspect of learning is signiﬁcant.

Participants reported that AI tools helped them un-

derstand new technologies and unfamiliar stacks, as

observed by P46 (Developer): “The role of AI in in-

creasing my learning process in these stacks has been

fundamental for my delivery deadlines and conven-

tional quality of the codes I produce.” This indicates

that AI tools can serve as virtual mentors, assisting in

professional growth.

The automation of repetitive tasks was another key

beneﬁt identiﬁed. As mentioned by P27 (Developer):

“This is laborious work that takes away useful time

and energy for solving more complex and speciﬁc

problems.” By automating these tasks, AI tools al-

low developers to focus their efforts on more complex

challenges, potentially increasing innovation and the

quality of the solutions developed.

5.2 Challenges and Limitations of AI

Tools

Despite the beneﬁts, participants also pointed out

challenges and limitations. The need for constant re-

view of the generated code was highlighted, with con-

cerns about “hallucinations” of the tools, where the

suggested code may not be accurate or appropriate.

P8 (Developer) warned: “There would always be su-

pervision due to common hallucinations in free AI

versions” This emphasizes the need for developers to

maintain an active and critical role when using these

tools, not blindly trusting the suggestions provided.

Additionally, some participants noted a lack of

complete understanding of the project’s context by AI

tools, as mentioned by P25 (Product Owner). This

limitation can lead to code suggestions that do not

perfectly align with the project’s speciﬁc require-

ments.

A point of concern raised was the risk of exces-

sive dependence on AI tools. P30 (Developer) ob-

served: “People have become dependent on them to

the point of not wanting to think anymore and go-

ing straight to consult them.” This observation raises

questions about the long-term impact on the develop-

ment of professionals’ skills. If developers rely too

much on AI tools, this can negatively affect their abil-

ity to solve problems independently and think criti-

cally.

6 LIMITATIONS AND THREATS

TO VALIDITY

The primary limitation of this study is its focus on

a single organization, which may limit the general-

izability of the ﬁndings to other contexts or indus-

tries. Additionally, as AI tools continue to evolve

rapidly, the relevance of the results could diminish

over time, necessitating ongoing research. This study

provides an observational perspective on the use of AI

tools within the organization, offering insights rather

than establishing cause-and-effect relationships. Fu-

ture studies could expand the scope to include multi-

ple organizations and explore performance metrics to

complement these ﬁndings.

The Impact of AI Tools on Software Development: A Case Study with GitHub Copilot and Other AI Assistants

251

7 CONCLUSIONS

In this study, we investigated the impact of using Ar-

tiﬁcial Intelligence (AI) tools, such as GitHub Copi-

lot, ChatGPT, and Gemini, in software development

within an R&D organization. Through quantitative

and qualitative analysis, we sought to understand how

these tools inﬂuence productivity, code quality, job

satisfaction, and other relevant aspects for profession-

als involved in the software development cycle.

The quantitative results indicated a signiﬁcant

adoption of these tools, with most professionals us-

ing them daily or several times a week. The perceived

beneﬁts include a substantial increase in productivity,

reduced time on repetitive tasks, improved code qual-

ity, and acceleration in the learning and knowledge

acquisition process. In addition, many participants re-

ported that AI tools helped them meet deadlines more

efﬁciently and focus on more complex tasks, delegat-

ing routine tasks to intelligent assistants.

The qualitative analysis complemented these ﬁnd-

ings, revealing that developers value the ability of AI

tools to generate and search code quickly, improve

code quality through efﬁcient suggestions, and facili-

tate learning of new technologies. The positive impact

on documentation, proposal preparation, and obtain-

ing insights and mentoring was also highlighted.

However, participants also identiﬁed challenges

and limitations. The need for constant review of the

generated code, the ’hallucinations’ of the tools, and

the lack of complete understanding of the context

were pointed out as points of attention. There were

concerns about the possible excessive dependence on

AI tools, which could affect developers’ ability to

think critically and solve problems autonomously. In

addition, the importance of organizational support,

especially concerning the acquisition of paid licenses,

could mitigate some of the observed disadvantages.

While this study evaluated the impact of AI tools,

a comparative analysis of Gemini, ChatGPT, and

GitHub Copilot remains an opportunity for future re-

search to further explore their unique strengths and

applications in each of the software engineering roles.

ACKNOWLEDGEMENTS

We would like to thank Fundac¸

ao Paulo Feitoza

(FPFtech) for the ﬁnancial support, as well as the

opportunity to conduct this research and present it.

Additionally, we acknowledge all participants who

kindly contributed to this study. We also acknowledge

the use of artiﬁcial intelligence–generated text in the

preparation of this article. Speciﬁcally, the sections

Introduction and Discussion were produced with the

assistance of the language model ChatGPT.

REFERENCES

Imai, S. (2022). Is github copilot a substitute for human

pair-programming? an empirical study. In Proceed-

ings of the ACM/IEEE 44th International Conference

on Software Engineering: Companion Proceedings,

ICSE ’22, page 319–321, New York, NY, USA. As-

sociation for Computing Machinery.

Lim, W. M., Gunasekara, A., Pallant, J. L., Pallant, J. I.,

and Pechenkina, E. (2023). Generative ai and the

future of education: Ragnar

ok or reformation? a

paradoxical perspective from management educators.

The International Journal of Management Education,

21(2):100790.

Lira, W., Neto, P. S., and Osorio, L. (2024). Uma an

alise do

uso de ferramentas de gerac¸

ao de c

odigo por alunos de

computac¸

ao. In Anais do IV Simp

osio Brasileiro de

Educac¸

ao em Computac¸

ao, pages 63–71, Porto Ale-

gre, RS, Brasil. SBC.

Moon, J., Yang, R., Cha, S., and Kim, S. B. (2023). chat-

gpt vs mentor : Programming language learning as-

sistance system for beginners. In 2023 IEEE 8th In-

ternational Conference On Software Engineering and

Computer Systems (ICSECS), pages 106–110.

Peng, S., Kalliamvakou, E., Cihon, P., and Demirer, M.

(2023). The impact of ai on developer productivity:

Evidence from github copilot.

Rajlich, V. (2014). Software evolution and maintenance. In

Future of Software Engineering Proceedings, FOSE

2014, page 133–144, New York, NY, USA. Associa-

tion for Computing Machinery.

Yetis¸tiren, B.,

Ozsoy, I., Ayerdem, M., and T

un, E.

(2023). Evaluating the code quality of ai-assisted code

generation tools: An empirical study on github copi-

lot, amazon codewhisperer, and chatgpt.

Ziegler, A., Kalliamvakou, E., Li, X. A., Rice, A., Rifkin,

D., Simister, S., Sittampalam, G., and Aftandilian, E.

(2022). Productivity assessment of neural code com-

pletion. In Proceedings of the 6th ACM SIGPLAN

International Symposium on Machine Programming,

MAPS 2022, page 21–29, New York, NY, USA. As-

sociation for Computing Machinery.

ICEIS 2025 - 27th International Conference on Enterprise Information Systems

252