THE USE OF ARTIFICIAL INTELLIGENCE APPLICATIONS FOR EDUCATION AND SCIENTIFIC RESEARCH

Artificial Intelligence (AI), particularly Generative Artificial Intelligence (GAI), is increasingly being used in education and scientific research. This article explores the use of AI applications, such as ChatGPT, Scite

Palabras clave: Aplicaciones de Inteligencia Artificial, ChatGPT, Scite, Litmaps, investigación In its 2023 Future of Jobs report, the World Economic Forum (2023) stated that AI is expected to create 58 million new jobs by 2025, most of which will require some expertise in AI and machine learning.According to this report, by 2030, AI is expected to contribute US$ 15.7 trillion to the global GDP, a 26 % increase from the current figure.Approximately 40 % of this growth is attributed to enhanced productivity boosted by AI.The key takeaway is that for formal education to fulfill its commitment to actively empowering students to engage in society and contribute to its ongoing development, students must acquire knowledge, competence, and confidence in building and utilizing AI within the framework of their formal education.Thus, regardless of whether we see AI as an opportunity or, instead, as a threat, the education system will need to make AI a prominent element of our curricula.

Large Language Models (LLM) and Generative Artificial Intelligence (GAI)
New developments in natural language processing (NLP) have resulted in the creation of robust language models, including large language models (LLM), trained with massive amounts of data, that allow to generate human-quality text, translation of languages, writing different kinds of creative content, and answer your questions in an informative way.LLMs are still under development but have already been used to create various applications, including chatbots, virtual assistants, and content-creation tools (Liu et al., 2023).GPT-3/GPT-4, as an illustration, exhibits the ability to generate text with human-like qualities based on the input it is provided, thereby serving as a potent tool for diverse natural language processing tasks.In Generative AI (GAI), "generative" refers to these systems' ability to produce unique and diverse outputs beyond simply recognizing or categorizing preexisting data.This area of Artificial Intelligence is concerned with creating new content, frequently by identifying patterns and structures in alreadyexisting data.It entails developing algorithms and models to generate new, realistic data across various media formats, including text, images, audio, and other types.The Generative Pre-Trained Transformer (GPT) (Radford et al., 2018) comprises models like GPT-3/GPT-4, is a prime example of GAI.
Another example of LLM is Bard (Bard, Google AI, 2023).Bard has a range of capabilities, such as generating types of content like poems, code, scripts, musical pieces, emails, and letters.Additionally, Bard can translate text between several languages.It can also summarize texts such as articles or books and offer factual topic summaries.Moreover, Bard can convert text to speech for individuals with impairments or reading difficulties and translate speech into text for those who are deaf or have hearing challenges.This paper reviews the use of some GAI applications for scientific research and discusses its benefits and pitfalls, including ethical aspects that need to be addressed by the community.

MATERIALS AND METHODS
We used Google Scholar, ChatGPT, Scite, and Litmaps as the primary sources of references and data to write this paper.
The paper is structured in the following manner: In the following sections, we review ChaGPT as a powerful GAI system, Scite to explore trends in bibliographic references, and Litmaps for citation relationship networks.In the discussion, we will present their main strengths and threats in the research process, finally establishing the study's conclusions.

ChatGPT
ChatGPT (Chat Generative Pre-Trained Transformer) is an artificial intelligence-based conversational large language model (LLM) developed by OpenAI (Radford et al., 2018;OpenAI, 2023).It is a chatbot with a conversational artificial intelligence interface that has gained prominence in various domains, including medical education and healthcare literature (Sallam, 2023a).ChatGPT can generate text responses from its knowledge acquired via machine learning in engagement with data from large databases and the internet (Pavlik, 2023).As an example, Figure 1 shows a ChatGPT answer to a question regarding photosynthesis.

ChatGPT in health
In healthcare practice, ChatGPT has been identified as a valuable tool for streamlining workflow and refining personalized medicine (Sallam, 2023a).
ChatGPT can provide information from updated evidence-based guidelines, making it a valuable tool to augment a physician's clinical judgment (Ghafoor & Yahya, 2023).Furthermore, the utility of ChatGPT in specific domains such as cancer treatment information, infectious disease research, and pharmacy has been explored, indicating the growing popularity of large language models for medical question-answering and information retrieval (Chen et al., 2023;Cheng et al., 2023).
The performance of ChatGPT in clinical reasoning exams and its implications for medical curricula and practice have also been examined, suggesting the need to revise assessments and incorporate AIrelated topics into medical education (Strong et al., 2023).
However, there are concerns about the accuracy of the information it provides in the healthcare domain, particularly in generating patient education materials and answering specific medical queries (Sütcüoglu & Güler, 2023).Despite these concerns, ChatGPT can be used in medical, dental, pharmacy, and public health education to improve personalized learning and clinical reasoning (Sallam et al., 2023).

ChatGPT in Science and Education
Several studies have explored the potential applications, benefits, and concerns of integrating ChatGPT into various scientific research and academic writing aspects (Cheng et al., 2023;Deng & Lin, 2023;Sallam, 2023aSallam, , 2023b)).The potential benefits of ChatGPT in assisting non-native English users and its impact on productivity in higher education have also been highlighted by Lin, (2023), Firaina & Sulisworo, (2023) and, Rahman & Watanobe (2023) suggested it could help students improve their programming skills.
In the context of academic writing, ChatGPT speeds up scientific writing.It is an efficient language bot that generates text by linguistic connections (Buriak et al., 2023), which is helpful for second language writing (Chen, 2023).Moreover, Seth et al. (2023) highlighted it as a tool that could revolutionize academia and increase research writing.

Issues and limitations of ChatGPT
The scientific community raised concerns about using ChatGPT and other AI applications in education, particularly regarding the breakdown of academic integrity and the lack of ethics involved in copying and plagiarism (Khalil & Er, 2023;Mohammadzadeh et al., 2023).On the other hand, there are concerns regarding the impact of ChatGPT on the quality, authenticity, and critical thinking in scientific research and academic writing (Arif et al., 2023;Rahman et al., 2023;Sallam et al., 2023;Seth et al., 2023).
The ChatGPT-4 fails to generate accurate scientific references in various domains, including Mechanical and Electrical Engineering (Barrot, 2023;Chen, 2023), and some health domains (Sütcüoglu & Güler, 2023).ChatGPT may produce false or incorrect references when queried on a scientific topic.As a result, text generation could be biased or inaccurate.For this reason, we advise against using the current versions of this application to generate high-level or in-depth scientific texts.
To overcome issues with generating proper citations of ChatGPT, other applications, such as Scite, are shown in the next section.

Scite
Scite is a platform designed to assist researchers in assessing the credibility of papers and their references.Scite, utilizing AI, classifies citations Wasi 3(1), 2024 ISSN: 2955 -8255 into three categories: supporting, contradicting, or simply mentioning the referenced research.This tool enables users to efficiently evaluate the context and significance of papers (Brody, 2021;Nicholson et al., 2021).The current implementation of Scite (Scite.ai)offers several tools for researchers, among them:

Hatun Yachay
• Scite Assistant, using a proprietary LLM and an extensive database of citations, gives researcher text generated based on references related to the user query, as shown in Figure 2.
• Search for scientific references in large databases and use the data to find references supporting or contradicting citations.• Verify the references' reliability of a paper to check if they have been retracted or supported by other studies.• Provide the researcher with a publication's dashboard with several valuable metrics, such as journals and authors citing the most from their work.

FIGURE 2 A ChatGPT answer to a question regarding photosynthesis
Note: There are two main windows: on the left, text generated, and on the right, a list with references used.
There are other features to find networks of references.However, Litmaps are the best option, as described in the following section.

DISCUSSION
This section summarizes the advantages and disadvantages of using ChatGPT in education and research and using Scite and Litmaps in scientific research.Finally, we reference the ethical considerations for using GAI in education and research.

ChatGPT in Education
ChatGPT offers advantages in the field of education: • Interactive learning.With its interface, ChatGPT allows students to engage in natural language conversations, making the learning process more engaging and enjoyable.• Personalized support.ChatGPT can assist students by answering questions, explaining concepts, and offering tailored explanations based on needs.• Availability.One of the benefits of ChatGPT is its round-the-clock availability, ensuring that students can access information and assistance at any time.This flexibility is particularly beneficial for learning and students residing in different countries.
• Language practice.For those learning a language, ChatGPT serves as a tool for practicing and improving language skills through interactive conversations.It can assist with grammar, vocabulary, and overall comprehension.To make the most of ChatGPT in research, researchers must approach its use with caution.
Recognizing its limitations and considering it as a tool rather than a replacement for human expertise and critical thinking are essential factors.Human oversight and verification play roles in ensuring rigorous scientific inquiry.

Scite in research
Scite.ai provides benefits in the context of research: • Scite.aispecializes in analyzing citations, allowing researchers to evaluate the impact and credibility of a scientific paper by examining its citation context.This feature assists researchers in understanding how a specific work has been received and referenced within the community.It is important for researchers using Scite.aithat while it provides features, they should also consult sources of information and apply critical thinking when evaluating research papers.Additionally, users should be mindful of any biases in the data and algorithms used by Scite.ai.
Although Scite.ai provides useful features, its usage in scientific research is limited by drawbacks.
• The accuracy and dependability of Scite.aisanalyses hinge on the quality of the underlying data, including the precision of citation information.Any inaccuracies or missing data might impact the reliability of the results.Organizations and users should adhere to comprehensive ethical use policies to ensure the use of GAI models.These policies should prioritize integrity, discourage plagiarism, and provide guidelines for responsibly utilizing AI-generated content.
Academic institutions must inform researchers and students about the potential hazards of using GAI models while highlighting the importance of producing work.Guidelines for citation styles and techniques to prevent inadvertent plagiarism must be provided.Authors should share research data for result reproduction; it is essential for ensuring transparency and facilitating the verification of scientific findings.
We have primarily used AI to prepare class materials, summarize and simplify texts, and brainstorm ideas.
In our institutions, guidelines are in preparation.However, other universities have already defined their guiding principles (Moorhouse et al., 2023;Russell Group, 2023).

CONCLUSIONS
GAI applications such as ChatGPT or Scite in education and research have several advantages.However, they should be used with caution.The text generated may contain biases and inaccuracies from incomplete or biased data sources and references.Authors should only use currently available AI tools for knowledge inference and only assume conclusions to be valid after proper verification.
Current AI technologies, such as Scite and Litmaps, offer analysis and network representation of citations.However, results in several fields may be limited by the quality and completeness of databases.
Academic institutions should inform students and researchers about the hazards of using GAI models.Scientists should be provided with clear and proper ethical guidelines to follow.

FIGURE 1 A
FIGURE 1 A ChatGPT answer to a question regarding photosynthesis

Figure 3 ,
Figure 3, the relationships network of a research article are depicted with time (x-axis) and the relevance map of the article.
• Research aid.Students can rely on ChatGPT to quickly gather information on topics, helping them with their research endeavors.It provides summaries, explanations, and references to sources that accelerate the research process.• Coding support.In computer science and programming courses, ChatGPT proves helpful by offering code examples that explain programming concepts clearly and provide • Automated notetaking.Researchers can utilize ChatGPT to take notes during meetings or while THE USE OF ARTIFICIAL INTELLIGENCE Hatun Yachay Wasi 3(1), 2024 ISSN: 2955 -8255 • The Scite.ai database may not uniformly cover all journals or fields.Researchers involved in The text produced by GAI models may unintentionally generate content that looks much like previously published work, which could result in plagiarism claims.Although GAI models aid in text generation, users must guarantee their work's originality.Users must check ChatGPT's output and make any necessary modifications to ensure that it adheres to standards and upholds academic integrity.In research environments, peer review and human oversight are still fundamental.While GAI models can assist in generating content, human experts need to review and verify the generated materials' accuracy, originality, and appropriateness.
It is recommended that researchers view Litmaps as a tool rather than relying solely on its capabilities.Additionally, they should augment insights gained THE USE OF ARTIFICIAL INTELLIGENCEHatun YachayWasi 3(1), 2024 ISSN: 2955 -8255