ISSN 1390-7778 (Versión Impresa)

ISSN 2528-8148 (Versión Electrónica)

YACHANA

Revista CientífiCa

Volumen 13, Número 1, Enero-Junio 2024

Fecha de recepción:

Fecha de aprobación:

Fecha de publicación:

Esta publicación está bajo una

licencia Creative Commons

Atribución-NoComercial 4.0

Internacional (CC BY-NC 4.0).

Abstract

This article examines the incorporation of

articial intelligence (AI) in the research

and writing practices of Mexican scien-

tists. Semi-structured interviews were

conducted with researchers in the elds

of biology, biological sciences, and bio-

medical sciences. The most common use

of AI tools was found to be to improve,

transform, and adapt texts, especially in

correcting English manuscripts, thus fa-

cilitating publication in international con-

texts. Other uses include searches in the

literature, suppchatort as a virtual tutor,

preliminary data interpretation, and gener-

ation of programming code for statistical

analysis. Only one participant uses AI as a

research assistant to dene objectives and

methodologies. All participants agreed

that AI improves their efciency. How-

ever, they also identied signicant issues

such as plagiarism, information leakage,

and academic dishonesty. They also de-

nounced a practice they consider blatant:

people presenting complete AI-generated

texts as their own, a practice observed in

both students and established researchers.

Taking into account these challenges, it is

essential to implement training programs,

awareness campaigns, and establish up-

dated ethical guidelines. The article con-

cludes with recommendations to improve

transparency and declaration of AI use in

research.

Keywords: Scientic personnel train-

ing, Research strategies, Articial Intel-

ligence.

Resumen

Este artículo examina la incorporación de

la Inteligencia Articial (IA) en las prácti-

cas de investigación y escritura de cientí-

cos mexicanos. Se realizaron entrevistas

semiestructuradas a investigadores de las

áreas de biología, ciencias biológicas y

https://doi.org/10.62325/10.62325/yachana.v13.n2.2024.924

Artículo de

Investigación

25/04/2024

02/07/2023

31/07/2024

The incorporation of generative articial

intelligence in the research and writing

practices of Mexican scientists

Eduardo Santiago-Ruiz

La incorporación de la inteligencia articial generativa en las

prácticas de investigación y escritura de cientícos mexicanos

Universidad Pedagógica Nacional, Ciudad de México-México. eduardoatx9@gmail.com

https://orcid.org/0000-0001-9450-2885

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

ciencias biomédicas. Se encontró que el uso más común de herramientas de IA es para me-

jorar, transformar y adaptar textos, especialmente en la corrección de manuscritos en inglés,

facilitando así la publicación en contextos internacionales. Otros usos incluyen la búsqueda de

bibliografía, el apoyo como tutor virtual, la interpretación preliminar de datos y la generación

de código de programación para análisis estadísticos. Solo uno de los participantes utiliza la IA

como asistente de investigación para denir objetivos y metodologías. Todos los participantes

coincidieron en que la IA mejora su eciencia. Sin embargo, también identicaron problemas

signicativos, como el plagio, la ltración de información y la deshonestidad académica. Tam-

bién denunciaron un uso que consideran descarado: personas que presentan textos completos

generados por IA como propios, una práctica observada tanto en estudiantes como en investi-

gadores consolidados. Ante estos desafíos, es fundamental implementar programas de forma-

ción, campañas de concientización y establecer lineamientos éticos actualizados. El artículo

concluye con recomendaciones para mejorar la transparencia y la declaración del uso de IA en

la investigación.

Palabras clave: Formación de personal cientíco, Estrategias en la investigación, Inteligencia

Articial.

Introduction

In recent years, the integration of articial

intelligence (AI) in research has been a

constantly growing trend. In a pioneering

article, Hutson (2022) explained the poten-

tial applications of AI in various scientic

elds. What seemed like science ction at

that time, as it was within reach of only a

few, became a tangible reality in just one

year with the arrival of ChatGPT. This

chatbot, which only requires an Internet

connection, allowed many scientists who

had no prior contact with AI to begin in-

tegrating it into their writing and research

practices.

According to the World Economic Forum

(2023), this technology will bring pro-

found changes to the job market in just

ve years. And one of the professions with

a high probability of being ‘augmented’ is

precisely that of scientists, as it can signi-

cantly improve their work.

Although there is abundant literature on

the use of AI in science, most of them

are essays, editorials, and opinion pieces,

making empirical research highly neces-

sary (Sallam, 2023). Furthermore, most of

the research has focused on other regions,

leaving an important gap with respect to

what is happening in Latin America. In a

context where AI is already beginning to

modify the work of researchers, it is cru-

cial to understand the changes that are oc-

curring. This involves understanding not

only how they are adapting to this tech-

nology and how they are leveraging it but

also the problems they face. Therefore, the

purpose of this article is to examine how

Mexican researchers in the elds of biol-

ogy, biological sciences, and biomedical

sciences are incorporating AI into their

research and writing practices. This study

adopts a qualitative approach that allows

for a comprehensive exploration of the

modications that occur in this eld. The

ndings are expected to contribute to a

better understanding of how AI is incorpo-

rated into academic and scientic elds in

general.

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

Literature review

The history of AI can be traced back to

Alan Turing (1950) who in his article

Computing Machinery and Intelligence

discussed the possibility of thinking sys-

tems. However, it is only in recent times

that AI has experienced a true explosion

in its capabilities, with such signicant

growth that it is transforming not only its

immediate environment but also having

social, economic, and political impacts on

a global scale (Dwivedi et al., 2023). Gen-

erative AI, exemplied by platforms such

as ChatGPT, Copilot, or Gemini, is charac-

terized by three key attributes.

First, its ability to execute a wide range

of tasks; second, its capacity to produce

novel content similar to what a human

would create; and third, the incorporation

of user interfaces that allow intuitive inter-

actions through natural language (Briggs

& Kodnani, 2023). Undoubtedly, the most

well-known is ChatGPT, which is a Large

Language Model (LLM), that is, an archi-

tecture that allows natural language pro-

cessing. But there are many other tools that

enable the creation of audio, image, video

or programming code, with a wide range

of uses ranging from video game creation

to business or science (Gozalo-Brizuela &

Garrido-Merchán, 2023).

In the specialized literature, AI is mainly

mentioned as a support tool for writing. Its

primary uses are text editing, title sugges-

tion, translation, content synthesis, stylis-

tic renement, and the creation of formal

communication texts (Else, 2023; Golan

et al., 2023; Kim, 2023; Santiago-Ruiz,

2024). Another use is bibliographic anal-

ysis. Vincent (2023) explains that, histor-

ically, literature review has been carried

out using the technological tools of the

time, such as Index Medicus, PubMed,

and now AI. According to the author, these

tools will allow for more exhaustive and

in-depth searches: “AI-based systems will

not only access more articles, but they will

also automatically select the most relevant

and analyze their quality” (Vincent, 2023,

p. 1). AI can also be used in various appli-

cations ranging from generating research

ideas, receiving feedback on one’s own

work, and generating programming code

(Bom, 2023; Hutson, 2022). In general,

there is a consensus that AI could be a

powerful ally, allowing time savings, in-

creased efciency, and improved produc-

tivity.

Despite its great features, this technology

is not free from risks and problems. One

of the most debated topics is that of false

information, which stems from the propen-

sity of LLMs to generate hallucinations,

that is, responses that, although they sound

plausible, lack sense (OpenAI, 2022).

Hallucinations make it necessary to thor-

oughly review all ChatGPT responses (Sa-

bzalieva & Valentini, 2023). Additionally,

it causes texts created entirely with these

tools, such as case reports (Alkaissi & Mc-

Farlane, 2023), reviews, or complete arti-

cles (Huang & Tan, 2023) to be plagued

with inaccurate information or invented

citations.

One of the great shadows hanging over AI

is that it has been used to generate and pub-

lish hundreds of articles with the purpose

of increasing productivity or the number of

citations for certain researchers (Cabanac

& Labbé, 2021; Van Noorden, 2021). This

is a phenomenon that could be explained

due to the demanding model of publish or

perish. However, it should be noted that

the widespread enthusiasm for ChatGPT

has largely eclipsed this debate. However,

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

it is reasonable to assume that this prac-

tice will intensify with the adoption of

LLM, since it was previously carried out

with less sophisticated tools. AI has also

raised dilemmas regarding the attribution

of authorship, especially since ChatGPT

was included as a coauthor in some arti-

cles (Grimaldi & Ehrler, 2023; Lee, 2023;

Stokel-Walker, 2023).

In response, editorial policies have gradu-

ally begun to be established (Nature, 2023)

which, in general, have the following rules:

1) it is suggested to use as a writing assis-

tant, 2) LLMs cannot be authors or coau-

thors, 3) the use of AI must be documented

and made transparent, and 4) human au-

thors must be responsible for the text.

Some voices have pointed out the possi-

bility that the use of AI could negatively

affect analytical thinking, an aspect that

is especially relevant in the training of

new researchers (Hill-Yardin et al., 2023;

Marchandot et al., 2023). In this regard,

some adopt an optimistic perspective, sug-

gesting that groups with limited experi-

ence could benet tremendously from AI,

as it might stimulate a creative process in

a particular eld of interest and/or identify

gaps in the literature (Golan et al., 2023).

However, others express concern about

the possible negative impact on intellectu-

al development: “It can endanger students’

willingness to develop skills like writing

and researching, and, above all, a blind

usage of ChatGPT does not build critical

thinking and problem-solving skills, which

are essential for academic and lifelong suc-

cess” (Dwivedi et al., 2023, p. 27). As can

be seen, AI is a rapidly evolving technol-

ogy that presents signicant opportunities,

but also challenges for scientic research.

Methodology

Semi-structured interviews were conduct-

ed with ten scientists (Díaz-Bravo et al.,

2013). The sample was obtained using a

snowball technique. Participants had to

meet the following selection criteria: 1)

actively conduct research in the areas of

biology, biological sciences, and biomed-

ical sciences 2) have at least a doctorate

degree, and 3) use some AI tool for their

research or writing. The characteristics of

the participants in this investigation are de-

tailed in Table 1.

The interviews lasted between 18 minutes

and 1 hour. They were conducted online

via the Zoom platform during February,

March, and June 2024. An informed con-

sent letter was provided, and anonymity

was ensured. The interviews were tran-

scribed using the Word tool and corrected

by a human. Subsequently, they were pro-

cessed using a thematic analysis technique.

Results and discussion

General characteristics of AI usage

The rst contact with AI varied. In one

unique case, it arose from family recom-

mendations: “it was thanks to my husband,

who explained to me that they existed and

how I could use them” (Anabel). Some

others received advertisements in appli-

cations and social networks: “It appeared

when I opened the Edge browser” (Joana)

and “Facebook started to make a lot of pro-

paganda about this AI thing” (Nicolás). On

the other hand, for most participants, the

introduction to AI came from people with-

in their academic environment, such as

colleagues or students. This is the case for

Constantino: “it was through a colleague’s

recommendation”; for Samara: “one of

my colleagues was just going to work on

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

an articial intelligence project” and for

Sandra: “because students started talking a

lot, especially about ChatGPT”. This high-

lights the growing use of these technolo-

gies in academic and scientic settings.

It might be assumed that, given the highly

specialized nature of their research, par-

ticipants would opt to use paid versions of

these applications; however, this is not the

case. It is notable that all participants use

free versions, as Joana expressed emphat-

ically: “Yes, of course it’s free, as if I’m

going to pay”.

The ways in which prompts are written

vary considerably. Some provide detailed

instructions like Samara: “I need this mes-

sage to be very assertive to communicate

this need, but without ceasing to be friend-

ly”. Although most prefer to use direct and

concise instructions. A notable example is

Sandra’s method, who uses a single word:

“I always use ‘improve’, ‘improve’ colon

and then I input the text, it’s my favorite”.

From what the participants mentioned, it

can be inferred that complex prompts are

not necessarily required, but rather know-

ing when and how to apply them.

Writing assistant

The applications of AI as a writing assis-

tant are diverse and encompass aspects

such as style correction, translation, sum-

marization, and title creation, as well as

adapting a manuscript to different con-

texts. Undoubtedly, the most recurrent use

is to improve English style. Previously,

participants devoted a signicant amount

Table 1

Participants

Pseudonym Level Research line AI tools Time using AI

Anabel SNI 1 Placenta ChatGPT 8 months

Samara Ph.D. Breast cancer

ChatGPT

Consensus

1 year

Joana SNI 3 Vitamins Copilot 1 year

Sandra SNI 2 Infections ChatGPT 6 months

Nicolás SNI Candidate Infections ChatGPT 1 year

Constantino SNI 1

Adipose tissue DeepL

6 months

Aurora SNI Candidate Plant physiology ChatGPT 2 months

Javier SNI 1 Ecology

ChatGPT

Scispace

1 year

Ramiro SNI 1 Breast milk

Copilot

ChatGPT

6 months

Erik SNI 2 Biomedical sensors

ChatGPT

Writefull

8 months

Notes: (1) In Mexico, the National System of Researchers (SNI) is a program aimed at promoting the scientic, technologi-

cal, and humanistic research activities of outstanding professionals in various disciplines. It is granted by the National Com-

mission for Science, Humanities and Technology (CONAHCYT) and includes both a recognition and nancial incentive.

There are several levels, each more prestigious than the previous one: candidate, 1, 2, 3, and emeritus. (2) The research line

is expressed succinctly to protect anonymity.

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

of time and effort to correcting their man-

uscripts, often having to allocate part of

their research budgets to pay for special-

ized editorial correction: “paying for ed-

iting is extremely expensive” (Sandra).

Currently, in contrast, they entrust this

revision to AI: “most of the time we had

to send it to an editor for style correction,

and now there are entire paragraphs that

I put into the chat for style correction”

(Sandra) and later: “for the rst time with

the chat correction, we didn’t receive any

style correction remarks” (Sandra). The

results are impressive, and several partici-

pants claim that their works corrected with

this technique are already in the process

of publication: “last month, the rst arti-

cle where I use these tools was submitted,

and it has optimized the correction time”

(Constantino); “we have two [articles] cor-

rected by ChatGPT right now” (Sandra)

and: “I submitted two book chapters, and

it also helped me a lot with the English”

(Nicolás).

As high-level researchers who need to

communicate abroad, AI tools have prov-

en to be very useful, for example, during

presentations or conferences: “I had to in-

troduce a researcher, and although I knew

how to do it, I had many doubts. So, a

friend suggested that I ask ChatGPT for

help to create a presentation script” (Ana-

bel). It is also widely used for formal com-

munications with colleagues or editors:

“to write emails, to conrm that what I am

writing has the intended meaning, because

sometimes there can be certain colloqui-

alisms that we don’t necessarily manage”

(Samara).

Generative AI tools are also capable of

generating peritexts, which are texts that

complement the original. For example,

they can assist in creating titles: “this past

week, I used it for titles, I tell it what I want,

and it creates such bombastic titles for me”

(Sandra). Titles are of great relevance as

they not only describe the research but also

need to be attractive and engaging, poten-

tially increasing the chances of gaining

more readership or funding. Similarly, it is

possible to ask the AI to adjust the size of

a text to a specic number of words. This

is useful, for example, to meet the strict

length requirements in research funding

applications: “I have a project I wrote, and

now I am submitting it for funding purpos-

es. So, to avoid rewriting everything with

a specic word count, I tell it to summa-

rize it in so many words” (Joana). It is even

possible to completely adapt the tone of a

text for dissemination purposes:

Yesterday, we had to submit something

for outreach, which was a 100-word pie-

ce. So, super family-friendly, talk about

our research projects. So, I went straight

to ChatGPT and said, look, help me write

a super-friendly text. I mean, I wrote my

awful text, we work with this, with that,

make it nice for me. And so, it made it for

me from the start, like, let me invite you to

see my lab (Sandra).

The researchers mentioned in the previ-

ous paragraphs use general tools such as

ChatGPT or Copilot. On the other hand,

Erik is the only one using a specialized

AI, Writefull, to prepare his articles before

publication. This platform is designed to

correct articles that are already nished or

in an advanced draft stage, with the main

function of perfecting them before submis-

sion.

It checks the classic structure of the article

if it has the title, the abstract, correspon-

ding authors, keywords, references, con-

ict of interest statement, all these major

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

headings that are usual in an article. Basi-

cally, they tell you if you are missing any

section, if it meets editorial criteria, how

the gures are, and if they are mentioned.

Some even reach such renement that they

tell you, hey, you have gure number 5,

but it is not mentioned in the text. That

part of more nitty-gritty work is what can

suddenly simplify for you (Erik).

Most participants perceive the use of AI as

a writing assistant as legitimate, similar to

using integrated tools in word processors:

“for me, it’s like using the tool that comes

in Word to check my spelling” (Joana).

Several participants (Nicolás, Constanti-

no, Ramiro, and Erik) state that using AI

in this way brings them peace and calm, as

it helps them overcome their difculties in

writing in English. By acting as a writing

assistant, AI allows them to maintain full

control over their research; essentially, it

remains their results and ideas but adapted

to another language, for specic platforms

or target audiences. In this sense, AI can

be seen as a futuristic typewriter capable

of transforming text in a complex and pro-

found way.

Research assistant

In this context, AI is used to help dene

objectives, formulate hypotheses, design

methodologies, and make corrections

suggested by reviewers. Javier is the only

participant who uses AI in this way. He

explains that one of his articles had been

rejected multiple times over the past few

years, and due to all the changes suggest-

ed by the reviewers, it had lost its original

essence. So, he turned to ChatGPT for sug-

gestions on how to recover certain points

of his text:

It suggested that in the introduction I could

say this, in the methods I should say that,

in the results I should state this, and in the

discussion, I should mention that. From

this analysis, it suggested a couple of pa-

ragraphs for each section. We did recover

some ideas from those paragraphs, espe-

cially the hypothesis, as it was well writ-

ten. I felt it captured what I wanted to say

(Javier).

Javier took a free course offered in the

Facebook groups of Mexican researchers

belonging to the SNI. This course prom-

ises to teach how to “train” ChatGPT to

leverage it for research, speed up litera-

ture searches, and adapt it to one’s writing

style. Based on Facebook comments from

the person offering the course, ChatGPT

Plus and plugins such as Zapier are used.

According to Javier, he learned the follow-

ing aspects in the course:

You can teach it about your topics. For

example, you can tell it, I want you to learn

about this topic, what these three authors

say. The denitions given by such and

such on this topic. And then, based on what

I taught you, help me write an objective or

a research question or a method. Or from

this article where this method was used,

help me write how I can use this method in

such a situation (Javier).

Javier’s feelings about this are ambiguous.

He claims that he maintains control of his

research since he decides which sugges-

tions to apply, but he also has some doubts.

For instance, the coauthor of the article, his

graduate student, openly expressed feeling

uncomfortable using AI in this way:

He told me that he would feel bad for fo-

llowing its advice. And I told him, it’s not

that it’s giving us the idea, but it’s working

from the idea I gave it. We didn’t ask it

to solve this article that has been rejected

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

1,000 times. No, we said, we want it to say

this, help us say this (Javier).

In another part of the interview, Javier

hints at a feeling of guilt: “a certain guilt or

I don’t know what else to call it” (Javier).

He reects on the dangers of abusing this

technology: “At what point do you stop

being the one writing? So far, I feel like

it’s been me, but I think it’s very easy to

eventually leave it all to it” (Javier).

Searching, explaining, analyzing: new

ways to access knowledge

AI is also being integrated into various as-

pects of research practices. Its applications

are diverse, and here they are classied

into three main groups: literature search,

resolving queries (akin to tutoring), and

data analysis.

The rst of these, literature search, is a

strategy employed by Samara, Joana, Javi-

er, Ramiro, and Nicolás. For example,

Nicolás explains that he needed to know

when a particular technique was rst used,

and ChatGPT allowed him to nd the orig-

inal source: “It told me that such a tech-

nique for the classication of E. coli was

by so-and-so in such a year, and then of

course I went to PubMed. So, doing the

search there, I found the article” (Nicolás).

This researcher used the clues provided by

ChatGPT to then search for a specialized

database. However, other tools facilitate

direct access to government or research

documents. Such is the case with Copilot:

“it gives you the reference, you click on

the little number, and it takes you there, all

very easy to use” (Joana). Another similar

application is Consensus, which also al-

lows access to specialized references:

Consensus takes a screenshot of the sec-

tion containing the information you are

looking for, so instead of reading the entire

article or all the articles and then following

the thread, it helps you focus your search

much better (Samara).

Samara stands out as the most experienced

user in this regard, as she can use ChatGPT

as her rst tool to better understand a top-

ic and then access specic documents

through Consensus.

AI can also function as a kind of tutor to

resolve doubts or explore topics they do

not fully master (Joana, Sandra, Samara,

and Nicolás). The queries of these par-

ticipants cover methodologies, laboratory

techniques, and the use of specialized soft-

ware. For example, Joana explains:

What would I have done on other occa-

sions? I would have had to go to Maniatis.

And now look for the buffer, now look for

the other buffer, I would have taken 3 cen-

turies longer and then once I found both,

see how they differ, what the differences

are. I would have had to do it manually, it

would have taken much longer, maybe 1

hour, and here it took me 3 minutes (Joa-

na).

On the other hand, Sandra asks one of her

students to consult ChatGPT about how to

perform certain actions in SPSS: “Before

we had our friend [ChatGPT], we used to

watch YouTube videos, but it was very

time-consuming. So, the chat says that it

is in the le tab, and then quickly he gets

it” (Sandra). And Nicolás uses AI to advise

him on using Matlab and Cytoscape:

ChatGPT says to load such a library un-

der such conditions to get the network, or

search for metabolism genes, search for

such and such, so you can build it. Then it

gives your ideas for building the network

or things related to expression, things that

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

I would never have imagined. And then it

gives you an idea, and you go and search

for more (Nicolás).

The interviewees are immersed in an envi-

ronment that requires the constant acqui-

sition of new laboratory techniques and

software skills. This demand drives them

to seek new sources of knowledge to stay

updated: “I always tell [my students], just

take a course, and they say no, the chat

helps me” (Sandra). In this context, AI is

partially replacing traditional ways of re-

solving queries and obtaining continuous

education, such as search engines, tutori-

als, videos, or courses, due to its efciency,

speed, and ability to personalize responses.

The nal use of AI observed among par-

ticipants is for data analysis and process-

ing. Nicolás takes his databases and inputs

them into ChatGPT for a preliminary anal-

ysis. Samara uses the same tool as a rst

approach to interpreting her results.

I ran some tests to see if a group of pro-

teins had correlations. You can do this in

a database. But often, most researchers do

not necessarily know how to handle data-

bases. So, I think ChatGPT is an easy way

to approach these things before going to

the database to understand it.

I simply asked it which process these pro-

teins might connect in, and I provided the

list. And then it told me, this protein does

this, and this protein does that. And then I

said, oh, okay, if both are involved in gly-

colysis, then it’s through metabolism that

they are connected (Samara).

Its use in statistical analysis can go even

further. For example, Ramiro uses it to de-

termine “the best statistics for such study

groups.” Nicolás generates code in the sta-

tistical analysis language R, in a process

that involves searching for existing code,

adapting it to his needs, and rening it:

What I do is download R codes, and ba-

sed on what I’m asking it, I say, hey, make

me a pool between this code and that code

because what I need is a scatter plot. So, it

pools the codes and generates a new code.

Of course, you have to adjust it, but Chat-

GPT does the bulk of the work (Nicolás).

From these observations, it can be con-

cluded that articial intelligence has a very

diverse use in scientic research. Its appli-

cations range from learning new software

to improving statistical analyzes and are

summarized in Table 2.

From Efciency to taboo: perceptions

on AI

All participants highlight as an advantage

of AI its ability to increase efciency: “I

think the main benet is speed, and at the

same time, providing you with a text that

reads smoothly” (Anabel). This is un-

doubtedly why they continue to actively

use these tools in their research. However,

along with these advantages mentioned,

certain problems also arise. These prob-

lems can be divided into two groups: those

with identied solutions for which mitiga-

tion strategies have been developed and

those that are more complex and profound

and still lack solutions.

Within the rst group of problems, the one

that stands out most among participants is

the lack of accuracy in information, also

known as hallucination: “Not everything

that ChatGPT outputs will be accurate, so

I have to verify it” (Samara). This risk is

present, but nearly harmless, because all

participants conduct thorough verication

of the answers they receive. The next per-

ceived risk is data leakage: “My main fear

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

is data theft, that the data end up being

somehow identied by other groups who

could beat me to publication” (Anabel).

To combat this latter problem, participants

have developed the strategy of modifying

or removing sensitive information before

inputting it into AI platforms: “What I

do is include a paragraph where there’s

no personal or private information at all”

(Samara). Unlike other groups, such as

university students (Santiago-Ruiz, 2023),

participants are informed about the risks

associated with the use of this technolo-

gy, enabling them to develop strategies to

counteract them.

On the other hand, there are also two prob-

lems that they are unsure of how to con-

front. The rst of these is inadvertently

using information belonging to others: “It

can even give you information that you

shouldn’t use. For example, if it’s some-

one else’s original idea” (Joana). This is a

serious issue related to a signicant lack

of clarity regarding copyright and AI. The

second problem is what Joana and Nicolás

describe as the ‘blatant’ use of LLMs to

create repetitive texts lacking analysis.

Joana explains: “I realize that [a student]

simply asked ChatGPT or who knows

whom ‘translate this for me’ and put it as

is.” And Sandra adds: “I worry, for exam-

ple, seeing very well-written texts, and it’s

evident during seminars that they don’t un-

derstand anything.”

Several participants (Joana, Sandra,

Nicolás, Aurora, and Ramiro) have en-

countered the challenge of academic dis-

honesty. However, none of them has come

to a clear solution on how to address it:

“Before, we used to have them read and

do exercises. Now it is getting more dif-

cult because they can even ask the chat

to summarize a reading for them, so it is

complicated (Sandra). Ramiro explains

that he encourages its use only among

his graduate students, but not among un-

Table 2

Uses of AI in research

General Use Specic Use

Writing Assistant

• Writing Improvement

• Translation

• Generation of peritexts (titles, abstracts)

• Adjusting length

• Adapting to another context (scientic dissemination or creating a report from

an article)

Information Search

• General information

• Panoramic bibliography search (when was the rst time it was reported…).

• Specic bibliography search (give me references about…)

Research Assistant

• Objectives

• Hypotheses

• Methodology

• Applying reviewers' corrections

Data Analysis

• Preliminary data analysis

• Preliminary statistical analysis

• Assistant in selecting statistical tests

• Programming code (R, Matlab)

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

dergraduates because “they are not really

trying to enhance their research skills or

information integration”.

This problem occurs not only among

students, but also among established re-

searchers. Joana recounts her experience

as a reviewer of a paper:

In this case, it was a journal with a very

high impact factor, above 5, and I realized

that AI had been used blatantly. So I re-

plied to the editor and the author and told

them that the writing strongly suggested

AI had been used, and I wanted them to

disclose it, to say it. Then the author was

very honest and said ‘yes, indeed, we used

an articial intelligence tool to write it,’

but the editor didn’t care and published it

anyway. I don’t think even the disclosure

was made. So I think that’s wrong, really,

because the article clearly showed signs of

that. For example, it was very repetitive,

redened things that were already dened,

repeated certain ideas but written in a di-

fferent way (Joana).

This is something researchers consider to

be one of the despicable uses of AI. Joana

says: “it’s really a plague, a very nefari-

ous thing, plagiarism or copy-paste, or

that your mind no longer works, but it’s

all through a robot.” Nicolás shares sim-

ilar sentiments: “they don’t understand

what it’s doing, therefore, it’s not making

their work easier, it’s doing the work for

them.” Unfortunately, the blatant use of AI

is a reality and undoubtedly poses the risk

of replacing the thinking capacity of both

young and established researchers.

Despite using AI tools in several of their

publications, none of the participants have

declared their use. This is in part due to

the fear that editors would reject their pa-

pers. Additionally, the participants argue

that there are no clear criteria for making

this declaration: “Well, yes, how do you

declare it? That is when you think about

what concepts the editors have about this

because that’s entering a void” (Javier).

Erik, as an organizer of a conference, em-

phasizes the importance of declaring the

use of AI and following the lead of major

scientic publishers adopting a transparen-

cy policy.

We are required to run articles through pla-

giarism detection software, but some now

also detect the use of AI tools. There was a

case where the software alerted us, saying

hey, be careful, because it looks like these

types of tools were used here. But since we

do not have a declaration of whether the-

se tools were used or not, we could not do

anything about it. That’s why we’re mo-

ving towards clearly stating the rules to tell

you what can and cannot be done (Erik).

This variability, between being extremely

efcient and dishonesty, leaves partici-

pants uncertain about AI. On the one hand,

they recognize its importance, but also

view it with reservations: “it seems like

we all hide it, it’s like our toxic boyfriend”

(Sandra). While AI is already being active-

ly used in scientic environments, a taboo

is also emerging. A taboo related to fear of

plagiarism and blatant use, which greatly

hinders open dialogue about this technol-

ogy.

Conclusions

AI and LLM have a wide range of practical

applications and provide signicant bene-

ts to researchers. All participants strongly

agreed that using these tools helped them

save time and become more efcient. AI

particularly stands out as a writing assis-

tant, which is expected to some extent,

given that LLM are designed to understand

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

and process natural language. In this area,

participants leverage AI through a variety

of strategies. For instance, to generate spe-

cic sections of a document such as titles

and abstracts. Moreover, it is commonly

used to adapt a text to different contexts,

such as re-ne a project into an article or

scientic outreach material. However, the

most frequently mentioned application is

manuscript editing in English. In this re-

gard, AI facilitates the integration of Mexi-

can scientists into an international research

environment and offers greater opportuni-

ties for publication in high impact journals.

AI can also be extremely useful in other

phases of research. For example, it can

assist in rening objectives or selecting

methodology. It should be noted that its

use as a research assistant is less wide-

spread and more controversial. Additional-

ly, AI is immensely valuable for informa-

tion retrieval, especially in a context where

the production of scientic literature is

overwhelming. Therefore, it would not be

surprising if future standards for review ar-

ticles were modied. Currently, PRISMA

incorporates specic aspects of search en-

gines, such as keyword usage. Thus, it is

plausible that future guidelines will be up-

dated to include features this technology.

The eld of AI tools is evolving rapidly.

While ChatGPT is the most well-known

application, there are many others. Copilot

is general-purpose; DeepL and Scispace

specialize in paraphrasing and text correc-

tion; Consensus is designed for bibliogra-

phy management, and Writeall focuses on

manuscript preparation before submission.

Each tool has specic capabilities and lim-

itations. It seems that this proliferation of

AI tools will only increase in the future.

It may not be sufcient to know just one

tool; rather, it will be necessary to use sev-

eral depending on their utility in different

research phases.

Despite the signicant advantages that AI

can offer, this technology also brings po-

tential problems such as plagiarism and in-

formation leakage. However, the greatest

perceived risk by participants is what they

have termed blatant use” This refers to the

practice of presenting entire texts or exten-

sive text sections generated by LLMs as

one’s own work. The result is document-

ing lacking depth, analysis, and creativi-

ty. This phenomenon appears to manifest

at all levels, from students to researchers

publishing in high-impact journals.

Recommendations to improve trans-

parency

It is crucial to explicitly disclose the use

of AI, as transparency is one of the cor-

nerstones of scientic work. Currently, al-

though AI is widely used in research, there

is a signicant regulatory gap. Therefore,

it is necessary for educational institutions,

publishers, and specialized journals to es-

tablish clear guidelines regarding the use

of AI. Furthermore, it is essential for the

authors to declare how they employ these

tools. So far, none of the participants

have declared their use of AI, sometimes

because they have not been explicitly re-

quired to do so, and other times due to fear

of editors’ opinions. Moreover, there are

no clear guidelines on how to explain the

use of AI. Therefore, the following recom-

mendations are provided to enhance trans-

parency.

• Explicitly state the AI tools used. Each

tool is designed for specic purpos-

es and has different capabilities and

limitations. Therefore, understanding

these characteristics is important to

evaluate their impact on research.

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

• Detail the application method. Provide

a detailed explanation of how AI tools

were used, highlighting the research

phases or the sections of the manu-

script where they were implemented.

This provides a clear context and en-

ables readers to understand the role AI

played in the process.

• Explain the prompts. Prompts can sig-

nicantly inuence results, so it is ad-

visable to describe them. This will help

to reproduce and evaluate the research

more effectively.

Based on the ndings of this research, it

can be concluded that the extensive func-

tionality of AI suggests that it will increas-

ingly take center stage in research in the

future. The changes that AI is bringing to

academic elds are substantial and alter

the way information is accessed, analyzed,

and written. Researchers who master these

technologies undoubtedly have a signif-

icant advantage over those who do not.

However, all of this is hindered by issues

such as blatant use and new forms of ac-

ademic dishonesty. Analytical thinking,

creativity, and transparency are essential

components of scientic work. However,

indiscriminate use of AI jeopardizes these

principles. This undoubtedly threatens the

development of new researchers, as sever-

al participants point out that many opt for

the easy path instead of developing their

cognitive abilities. Additionally, in today’s

context, which demands constant article

production and where high citation num-

bers are crucial, excessive use of AI could

undermine the scientic publishing sys-

tem. Currently, it is important to promote

positive uses of AI while limiting nega-

tives. This will not happen spontaneously

but requires us to create the right condi-

tions to achieve the best results. Therefore,

it would be necessary to overcome the

taboo and implement training programs,

awareness campaigns, and updated ethical

standards.

Declaration on the use of AI

The ideas, arguments, conceptual deni-

tions, research design, and interpretation

of the data are original creations of the

author. ChatGPT was used to accelerate

the writing process and correct Spanish

drafts with the prompt improve the writ-

ing. Once a nal version of the manuscript

was reached, ChatGPT was used with the

prompt translate to English. The author re-

viewed this translation to ensure coherence

and accuracy.

References

Alkaissi, H., & McFarlane, S. I. (2023,

February 19). Articial Hallucinations

in ChatGPT: Implications in Scientif-

ic Writing [Editorial]. Cureus, 15(2).

https://doi.org/10.7759/cureus.35179

Bom, HS. H. (2023). Exploring the Oppor-

tunities and Challenges of ChatGPT in

Academic Writing: a Roundtable Dis-

cussion. Nuclear Medicine and Molec-

ular Imaging, 57, 165–167. https://doi.

org/10.1007/s13139-023-00809-2

Briggs, J., & Kodnani, D. (2023, March

24). The Potentially Large Effects of

Articial Intelligence on Economic

Growth (Briggs/Kodnani). Goldman

Sachs. https://tinyurl.com/akjpmp3w

Cabanac, G., & Labbé, C. (2021, Decem-

ber). Prevalence of nonsensical algo-

rithmically generated papers in the

scientic literature. Journal of the

Association for Information Science

and Technology, 72(12), 1461–1476.

https://doi.org/10.1002/asi.24495

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

Díaz-Bravo, L., Torruco-García, U., Mar-

tínez-Hernández, M., & Varela- Ruiz,

M. (2013, julio-septiembre). La en-

trevista, recurso exible y dinámico.

Investigación en Educación Médica,

2(7), 162–167. https://www.redalyc.

org/pdf/3497/349733228009.pdf

Dwivedi, Y. K., Kshetri, N., Hughes, L.,

Slade, E. L., Jeyaraj, A., Kar, A. K.,

Baabdullah, A. M., Koohang, A., Ra-

ghavan, V., Ahuja, M., Albanna, H.,

Albashrawi, M. A., Al-Busaidi, A. S.,

Balakrishnan, J., Barlette, Y., Basu,

S., Bose, I., Brooks, L., Buhalis, D.,

… Wright, R. (2023, August). Opin-

ion Paper: “So what if ChatGPT wrote

it?” Multidisciplinary perspectives on

opportunities, challenges and impli-

cations of generative conversational

AI for research, practice and policy.

International Journal of Information

Management, 71, 102642. https://doi.

org/10.1016/j.ijinfomgt.2023.102642

Else, H. (2023, January 12). Abstracts

written by ChatGPT fool scientists

[News]. Nature, 613. https://doi.

org/10.1038/d41586-023-00056-7

Golan, R., Reddy, R., Muthigi, A., & Ra-

masamy, R. (2023, June). Articial

intelligence in academic writing: a

paradigm-shifting technological ad-

vance. Nature Reviews Urology, 20,

327–328. https://doi.org/10.1038/

s41585-023-00746-x

Gozalo-Brizuela, R., & Garrido-Merchán,

E. C. (2023). A survey of Generative

AI Applications. ArXiv. http://arxiv.

org/abs/2306.02781

Grimaldi, G., & Ehrler, B. (2023, January

4). AI et al.: Machines Are About to

Change Scientic Publishing Forever.

ACS Energy Letters, 8(1), 878–880.

https://doi.org/10.1021/acsenergy-

lett.2c02828

Hill-Yardin, E. L., Hutchinson, M. R.,

Laycock, R., & Spencer, S. J. (2023,

May). A Chat(GPT) about the future of

scientic publishing. Brain, Behavior,

and Immunity, 110, 152–154. https://

doi.org/10.1016/j.bbi.2023.02.022

Huang, J., & Tan, M. (2023). The role of

ChatGPT in scientic communica-

tion: writing better scientic review

articles. American journal of cancer

research, 13(4), 1148–1154. https://

tinyurl.com/yc8xbbvj

Hutson, M. (2022, October 31). Could

AI help you to write your next paper

[Technology Future]. Nature, 611,

192–193. https://tinyurl.com/3tme3m-

Kim, S.-G. (2023, March 8). Using

ChatGPT for language editing in sci-

entic articles. Maxillofacial Plas-

tic and Reconstructive Surgery, 45,

Article 13. https://doi.org/10.1186/

s40902-023-00381-x

Lee, J. Y. (2023). Can an articial intelli-

gence chatbot be the author of a schol-

arly article? Science Editing, 10(1),

7-12. https://doi.org/10.6087/kcse.292

Marchandot, B., Matsushita, K., Carmona,

A., Trimaille, A., & Morel, O. (2023,

March). ChatGPT: The next frontier in

academic writing for cardiologists or

a pandora’s box of ethical dilemmas.

European Heart Journal Open, 3(2),

oead007. https://doi.org/10.1093/eh-

jopen/oead007

Nature. (2023, January 24). Tools such as

ChatGPT threaten transparent science.

YACHANA Revista Cientíca, vol. 13, núm. 2 (julio-diciembre de 2024), pp. 83-97

The incorporaTion of generaTive arTificial inTelligence in The The incorporaTion of generaTive arTificial inTelligence in The

research and wriTing pracTices of Mexican scienTisTsresearch and wriTing pracTices of Mexican scienTisTs

Santiago-Ruiz, E

Here are our ground rules for their use

[Editorial]. Nature, 613. https://doi.

org/10.1038/d41586-023-00191-1

OpenAI. (2022, January 27). Aligning lan-

guage models to follow instructions.

OpenaiAI. https://tinyurl.com/39n-

vh5r2

Sabzalieva, E., & Valentini, A. (2023).

ChatGPT and Articial Intelligence

in higher education Quick start guide.

UNESCO. https://tinyurl.com/ym9h-

k4y4

Sallam, M. (2023). ChatGPT Utility in

Healthcare Education, Research, and

Practice: Systematic Review on the

Promising Perspectives and Valid

Concerns. Healthcare, 11(6), Article

887. https://doi.org/10.3390/health-

care11060887

Santiago-Ruiz, E. (2023). Writing with

ChatGPT in a context of educational

inequality and digital divide. Inter-

national Journal of Education and

Development Using Information and

Communication Technology (IJE-

DICT), 19(3), 28–38. https://tinyurl.

com/5n9aybfr

Santiago-Ruiz, E. (2024, February). Au-

tomation of administrative texts and

formal communication with ChatGPT.

Transdigital, 5(9). https://tinyurl.com/

ys4vwv3r

Stokel-Walker, C. (2023, January 18).

ChatGPT listed as author on research

papers: many scientists disapprove

[News]. Nature, 613. https://doi.

org/10.1038/d41586-023-00107-z

Turing, A. M. (1950, October). Comput-

ing Machinery and Intelligence. Mind,

New Series, 59(236), 433–460. https://

www.jstor.org/stable/2251299

Van Noorden, R. (2021, May 28). Hun-

dreds of gibberish papers still lurk in

the scientic literature [News]. Nature,

594, 160–161. https://doi.org/10.1038/

d41586-021-01436-7

Vincent, J.-L. (2023). How articial intel-

ligence will affect the future of med-

ical publishing. Critical Care, 27,

Article 271. https://doi.org/10.1186/

s13054-023-04511-9

World Economic Forum. (2023, Sep-

tember). Jobs of Tomorrow: Large

Language Models and Jobs. https://

tinyurl.com/v32a3b7v

Para referenciar este artículo utilice el siguiente formato:

Santiago-Ruiz, E. (2024, julio/diciembre). The incorporation of generative articial intelligence in the

research and writing practices of Mexican scientists. YACHANA Revista Cientíca, 13(2), 83-97.

https://doi.org/10.62325/10.62325/yachana.v13.n2.2024.924