Clase 1: El uso de la Inteligencia Artificial (IA) en la Metodología de la Investigación

Impact of Artificial Intelligence on Academic Research

Introduction to the Session

• The speaker expresses gratitude for the invitation and offers to share a PowerPoint presentation.

• The session begins with thanks to attendees and organizers, indicating an interactive format where questions are encouraged.

Speaker's Background

• The speaker is a tenured professor at Carlos III University in Madrid, specializing in research methods.

• Their research interests include artificial intelligence, computational social sciences, communication, and journalism.

Course Overview

• The course aims to explore the impact of AI on academic life, focusing on improving research quality and productivity.

• Emphasis is placed on using tools like ChatGPT as research assistants while highlighting the necessity of human supervision over AI outputs.

Structure of the Presentation

• The course is divided into two sections:

• How AI can assist throughout various stages of research.

• Analyzing interview data using AI techniques.

Interactive Learning Environment

• Attendees are encouraged to ask questions or interrupt during the presentation for a more engaging discussion.

• A dedicated time will be allocated at the end for addressing any remaining queries from participants.

Utilizing ChatGPT in Research

First Section: ChatGPT's Role in Research Design

• Discussion will cover how ChatGPT can aid researchers from idea conception through execution and analysis phases.

• Key aspects include ethical considerations and publication processes relevant to high-impact international studies.

Second Section: Data Analysis Techniques

• Focus will shift towards analyzing qualitative data from interviews using AI tools like thematic analysis.

Publication Expectations

• Understanding what international journals expect from submissions will also be part of the curriculum.

Practical Applications of AI Tools

Recommendations for Engagement with AI Tools

Impact of Artificial Intelligence on Research Practices

Ethical Use of AI in Research

• The ethical use of platforms like AI is crucial, and many journals are currently addressing this issue.

• There is uncertainty among social sciences journals regarding the recommendations for researchers on using AI tools effectively.

Transparency in AI-Assisted Research

• Researchers must disclose the use of AI in their studies; it should not be used covertly.

• Personal transparency is emphasized; acknowledgments should include statements about AI assistance in research tasks.

Qualitative Research Enhancement with AI

• The session focuses on accelerating qualitative research through AI, particularly in designing and executing interviews.

• Emphasis is placed on qualitative methods, which are essential for understanding perceptions and behaviors.

Tools for Data Analysis

• GPT should be viewed as a personal assistant rather than just a text generator; it aids in resolving research queries and generating ideas.

• Various qualitative data collection techniques exist, including surveys and ethnography, but the focus will be on interviews.

Software Impact on Research Efficiency

• Other software tools like Atlas.ti can significantly aid qualitative analysis by summarizing materials and facilitating scientific writing.

The Role of Artificial Intelligence in Qualitative Research

Introduction to AI in Research

• The integration of artificial intelligence (AI) has led to a significant transformation in both qualitative and quantitative research methodologies.

• This section will explore how AI can enhance the process of conducting in-depth interviews, providing concrete examples of its application.

Benefits of Using AI

• AI significantly reduces the time required for tasks traditionally outsourced, such as transcribing interview material and translating languages.

• Previously, researchers manually transcribed interviews and translated them; now, platforms can automate these processes efficiently.

• Tools like Whisper or ChatGPT can produce accurate transcriptions within an hour, compared to manual efforts that could take much longer.

New Research Scenarios Enabled by AI

• AI facilitates the proposal of new research scenarios, enhancing productivity for researchers working on theses or master's projects.

• It opens avenues for generating new lines of inquiry at the intersection of traditional research methods and the impact of AI technologies.

Practical Application: Conducting Research with AI Assistance

• The discussion includes how tools like ChatGPT can assist in qualitative data analysis and automatic text coding for reliable content analysis.

• Understanding why qualitative research is essential is crucial as future studies will increasingly focus on the implications and effects of AI.

Navigating Research with AI

• Researchers are guided on how to conduct studies with AI assistance while being aware that technology may sometimes yield biased results.

• It's emphasized that researchers should not relinquish complete control to AI; human oversight is necessary to ensure accuracy and relevance.

Example Case Study: Impact on Journalism

• An example involves using AI to generate a general objective for studying its impact on journalism practices.

• Providing context is vital when engaging with AI; without it, results may be inappropriate or irrelevant.

• A specific objective generated by the AI focuses on evaluating how it transforms editorial processes regarding efficiency and journalistic quality.

Research Objectives and AI in Journalism

Aligning Research Goals with AI Insights

• The research team evaluates whether the objectives generated by artificial intelligence align with their overarching research goals, ensuring that they are not accepted blindly but critically assessed.

• The role of AI is emphasized as a supportive tool rather than a directive force, highlighting the importance of human oversight in guiding research efforts.

Defining General and Specific Objectives

• The general framework for the study focuses on the impact of artificial intelligence on journalism, which serves as a foundation for developing specific research objectives.

• Three secondary objectives are established:

• To analyze perceived benefits of AI in newsrooms, acknowledging potential biases in journalists' perceptions.

• To investigate challenges faced by journalists when adopting AI technologies.

• To evaluate the impact of AI on productivity and creativity within journalism.

Importance of Research Questions

• Clear alignment between research questions and previously defined objectives is crucial to ensure coherence throughout the study. This alignment helps avoid common pitfalls seen in academic submissions where objectives do not match inquiry questions.

• The necessity for human supervision over AI-generated questions is stressed to maintain relevance and accuracy concerning the study's aims. If discrepancies arise, adjustments must be made to ensure consistency.

Examples of Research Questions

• Two key research questions are proposed:

• What changes has artificial intelligence brought to decision-making processes within newsrooms?

• How has AI affected interactions between journalists and audiences? These inquiries directly relate to both general and specific objectives outlined earlier, ensuring comprehensive coverage of relevant topics within journalism studies.

Conclusion on Human Oversight

Research Questions and the Role of AI in Research

Importance of Well-Formulated Research Questions

• Once research questions are clearly defined and aligned with study objectives, they guide the research process effectively.

• Research questions must be relevant and realistic; interesting questions may not always be answerable with available data.

• It is crucial that research questions can be addressed with existing data to ensure meaningful results.

Ethical Considerations in AI Usage

• AI should never replace human judgment; it facilitates but does not substitute for human oversight in research.

• Researchers bear full responsibility for ethical considerations when using AI tools like GPT, as accountability lies with the human team, not the machine.

Defining Boundaries in Research

• It's essential to delimit the time and space of research to avoid overwhelming amounts of information.

• Human criteria are vital for determining how far AI-generated content should go; researchers must maintain control over their work.

Interview Structure and Methodology

Developing an Interview Guide

• After formulating research questions, creating a structured interview guide is essential for planning interviews effectively.

Types of Interviews

• Different types of interviews exist: structured, unstructured, and semi-structured. Structured interviews ensure consistency across responses.

Benefits of Structured Interviews

• Using structured interviews allows all participants to respond to the same set of questions, facilitating comparative analysis across diverse perspectives.

Challenges with Semi-Structured Interviews

• While semi-structured interviews can provide valuable insights, they may lead to inconsistencies that complicate data analysis.

The Role of Researchers in Guiding Interviews

Interview Planning and Structure

Importance of Structured Interview Questions

• The planning of an interview involves formulating various questions organized into sections that align with previously established research questions.

• There is no fixed number of questions in an interview; it varies based on the research needs and the interviewer’s expertise.

Sections of the Interview Guide

• Typically, three distinct sections are used in an interview guide to address different aspects of the research topic.

• Example sections include:

• Introduction to Artificial Intelligence (AI) in Journalism

• Effects of AI on daily workflow

• Journalists' perceptions regarding changes introduced by AI

Crafting Questions for Each Section

• In each section, questions should be designed to elicit detailed responses relevant to that specific area. The number of questions can vary based on experience.

• For instance, Section Two focuses on how AI facilitates journalists' daily tasks, requiring targeted inquiries about its impact.

Finalizing the Interview Guide

• The completed interview guide typically spans one page, detailing each section along with corresponding questions aimed at addressing the research queries.

• Once this guide is prepared, interviews can commence as it provides a structured approach to gather necessary data.

Examples of Interview Questions

• An example question from Section Two could be: "How has your workflow changed since using AI tools?" This assumes prior use of AI by respondents.

• Another potential question might explore any barriers faced when integrating AI into their work routines and how they overcame them.

Demographic Considerations in Interviews

Interview Guide Development and Sampling in Qualitative Research

Importance of Individual Background in Interviews

• The inclusion of individual economic backgrounds is essential for establishing demographic differences, which can significantly impact research outcomes.

• It is recommended to incorporate demographic background questions into the interview guide to better understand participants' perspectives.

Advantages of AI in Interview Guide Creation

• Utilizing artificial intelligence (AI) for creating interview guides offers benefits such as speed, personalization, and cohesion. However, human supervision is crucial to ensure accuracy and relevance.

• Researchers must verify that AI-generated suggestions align with their research objectives to avoid biases or misleading information.

Selecting Interview Participants

• After developing the interview guide, determining whom to interview becomes critical; diversity among participants enhances the richness of data collected.

• Aiming for a heterogeneous sample—comprising various genders, ages, and professional backgrounds—will yield more comprehensive insights during interviews.

Heterogeneity in Sample Selection

• The goal is to maximize participant diversity to enrich data quality; this includes selecting journalists from local, regional, and national outlets. Greater diversity leads to richer evidence and improved study outcomes.

• A heterogeneous sample ensures varied experiences are captured, enhancing the analysis phase of research. This approach allows researchers to draw more nuanced conclusions from their findings.

Determining Sample Size: Theoretical Saturation

• A common question arises regarding how many interviews are necessary for effective research; theoretical saturation serves as a guiding concept here. Researchers should continue interviewing until no new evidence emerges from additional sessions.

• If after several interviews (e.g., 20+) new insights still arise, further investigation is warranted; however, once saturation occurs—where no new information is gained—it may be time to conclude data collection efforts.

Recommended Sample Sizes for Publication

• While some researchers report achieving theoretical saturation with fewer than 10 interviews, high-level journals typically require at least 25 interviews for publication credibility and depth of analysis. This number can vary based on specific journal standards but serves as a general guideline based on experience in qualitative research contexts.

Types of Sampling Methods in Qualitative Research

• When it comes to sampling methods within qualitative studies focused on interviews:

• Convenience sampling is one commonly used method that prioritizes ease of access over representativeness.

Sampling Techniques in Qualitative Research

Intentional Sampling in Journalism Studies

• The selection of interviewees is based on their availability and willingness, focusing on individuals who meet specific criteria relevant to the study, such as journalists in Spain or Peru.

Snowball Sampling Methodology

• The snowball sampling technique involves one interviewee providing contacts for subsequent interviewees, facilitating access to hard-to-reach populations.

• This method is particularly useful when researchers aim to connect with individuals fitting certain profiles but lack direct access.

Convenience Sampling Usage

• In cases where access to specific journalist profiles is limited, convenience or intentional sampling methods are often employed.

• Snowball sampling becomes essential when targeting groups at risk of social exclusion or those difficult to reach due to their circumstances.

Crafting Interview Invitations

• An example email invitation for participation in a study on AI and journalism includes an introduction and a request for a 30-minute interview while ensuring confidentiality.

• It’s important to highlight potential benefits for the interviewee, such as contributing to a better understanding of their work.

Ethical Considerations in Interviews

• Providing informed consent is crucial; participants should understand the study's objectives and how their data will be used while ensuring confidentiality.

• Researchers must protect participants from potential repercussions related to sensitive topics discussed during interviews, emphasizing anonymity through pseudonyms.

• Maintaining confidentiality means that data collected will only be used within the academic research context, safeguarding participant identities throughout the process.

Voluntary Participation and Withdrawal Rights

• Participants should know they can withdraw from the study at any time without consequence; this aligns with ethical research practices.

Multilingual Research Applications

Transcribing and Analyzing Interviews with AI

The Role of AI in Transcription

• Researchers can communicate in their native languages, with tools like ChatGPT and Whisper facilitating automatic translation during interviews.

• In qualitative research, data is primarily text from interviews, which necessitates transcription for analysis.

• Traditionally, audio was transcribed manually using headphones; now, platforms like Whisper automate this process significantly.

• While Whisper provides a good transcription service, it requires human oversight to correct inaccuracies that arise from automated processes.

• Users are encouraged to compare manual transcription with Whisper's output to appreciate the efficiency gained through automation.

Analyzing Interview Data

• After obtaining transcripts, thematic analysis is performed to extract key themes and subthemes from interview data.

• Thematic analysis involves identifying recurring topics discussed by interviewees and supporting these findings with direct quotes or evidence from the text.

• AI can also analyze non-verbal communication using tools like Gemini, which assesses facial expressions and body language from video recordings of interviews.

Ethical Considerations in AI Research

• Transparency is crucial; participants must be informed about the use of AI throughout the research process.

Discussion on AI in Research

Ethical Considerations of AI as Authors

• The association suggests that AI should not be credited as an author or co-author in research articles, emphasizing that responsibility lies with human oversight rather than technology.

Guidelines for Researchers Using AI

• Researchers must disclose the use of AI in their studies and refrain from listing it as a co-author, adhering to most journal policies.

Data Protection and Confidentiality

• In Europe, new regulations are emerging regarding personal data handling; confidentiality is crucial to protect both evidence and interviewee identities during research.

Advantages of Using AI in Research

• Key benefits include accelerating the research process, where AI acts like a personal assistant, streamlining tasks from transcription to data analysis.

Challenges Associated with AI Utilization

• While AI enhances efficiency, it introduces challenges such as dependency on technology and potential biases in data analysis due to its programming by humans.

Biases and Human Oversight

Risks of Bias Introduction

• The reliance on AI can lead to biased outcomes if the algorithms focus solely on specific directions, potentially overlooking diverse perspectives essential for comprehensive research.

Importance of Human Supervision

• Continuous human oversight is critical; while AI improves processes, it cannot replace the researcher’s role or ethical responsibilities in conducting studies.

Conclusions and Future Directions

Summary of Key Points

• Although AI enhances research efficiency, it must always be used ethically under human supervision. It does not substitute for researchers but serves as a tool to aid them.

Questions Raised During Discussion

• Participants sought clarification on how AI analyzes interview results and whether preliminary project outlines could guide its responses effectively.

Role of Human Knowledge

Understanding the Role of AI in Research

Importance of Human Guidance in AI Utilization

• The speaker emphasizes that while AI can guide researchers, it is crucial for them to be well-trained to provide clear instructions regarding their research direction.

• It is noted that the competitive advantage of well-trained researchers will not diminish with AI; instead, it will amplify their capabilities.

Ethical Considerations in AI Usage

• A question arises about the ethical necessity of acknowledging AI resources used in research, which cannot be cited as traditional references.

• The speaker suggests including such acknowledgments in a specific section at the end of scientific articles, typically labeled "Acknowledgments."

Workshops and Training on Qualitative Research

• An inquiry is made about whether workshops will be provided to illustrate qualitative research methods using AI tools.

• The coordinator confirms that there are planned workshops focusing on both quantitative and qualitative aspects related to AI.

Validation Processes for Interview Guides

• A participant expresses gratitude and raises a concern regarding the validation process for interview guides created with AI assistance.

• It is clarified that if tools like ChatGPT are used, this should be indicated during expert reviews or validations.

Contextualizing AI Tools for Effective Use

• The importance of providing context when configuring tools like ChatGPT is discussed; more context leads to better results.

Understanding the Role of AI in Academic Research

The Value of AI in Thesis Development

• The speaker discusses the equivalence of academic value between theses developed with and without artificial intelligence, emphasizing that the quality is determined by adherence to research criteria rather than the tools used.

• Quality in research is not dictated by whether AI tools like GPT are used; it depends on how appropriately they are applied to enhance the thesis.

Ethical Use of AI Tools

• Concerns arise when students rely too heavily on AI for writing their theses, which can lead to issues if they do not engage with the material themselves.

• The importance of verifying information sourced from AI is highlighted; researchers must ensure that data comes from credible sources to avoid misinformation.

Perspectives on Research Integrity

• A distinction is made between work supported by AI and traditional research efforts, suggesting that personal engagement in research holds more value.

• Ethical considerations are paramount; AI should serve as an assistant rather than a primary driver in research processes.

Challenges in Interview-Based Research

• Anonymity concerns during interviews are raised, particularly regarding how to verify the authenticity of responses provided by participants.

• The necessity for transcriptions from interviews is discussed as a means to validate participant contributions and ensure accuracy.

Trust and Verification in Research Practices

• While transcriptions can help confirm interview authenticity, there remains uncertainty about their reliability since they could be fabricated or generated by tools like GPT.

• Researchers must cultivate trust in their methodologies while acknowledging that absolute certainty about participant engagement cannot be guaranteed.

Consent Requirements for Using AI

• Questions arise regarding informed consent when utilizing both traditional methods and AI technologies during interviews.

Inteligencia Artificial y Plagio

Diferenciación entre plataformas de IA y plagio

• La inteligencia artificial (IA) y la subida de contenido a plataformas como YouTube son conceptos distintos que deben diferenciarse.

• Según la experiencia del doctor, Turnitin no clasifica los textos generados por IA como plagio, ya que su función es comparar textos con fuentes existentes.

• Actualmente, Turnitin no puede detectar si un texto fue generado por GPT; su objetivo principal es verificar coincidencias con fuentes académicas.

Herramientas alternativas para detección de IA

• Existen otras herramientas de IA, como ZeroGPT, que permiten verificar si un texto ha sido escrito por GPT y calcular el índice de coincidencia.

Análisis Temático en Investigación

Introducción al análisis temático

• El análisis temático permite analizar las transcripciones de entrevistas para extraer resultados significativos en investigaciones.

• Esta técnica fue desarrollada por investigadoras australianas en psicología y se ha expandido a diversas disciplinas científicas debido a su facilidad de uso.

Proceso tradicional del análisis temático

• Tradicionalmente, el análisis se realizaba manualmente; el investigador leía varias veces las transcripciones para identificar temas principales.

• Este proceso manual es más lento y costoso comparado con métodos automatizados que facilitan un análisis más ágil.

Implementación del Análisis Temático Automatizado

Preparación para el análisis automatizado

• En la investigación se realizaron entrevistas y se formularon tres preguntas clave. Se transcribieron 60 páginas de datos para el análisis.

• Se decidió utilizar GPT para realizar un análisis temático basado en las mismas transcripciones utilizadas previamente en un estudio manual.

Protocolo de investigación con GPT

• Se estableció un protocolo donde se instruyó a GPT sobre cómo responder a las preguntas formuladas durante la investigación.

• La preparación incluyó limpiar los datos eliminando etiquetas HTML y organizando las entrevistas numeradas para facilitar la interacción con GPT.

Definición del proceso analítico

Analysis Process and Key Themes in Interview Responses

Overview of Analysis Steps

• The analysis process involves identifying recurring themes based on interviewee responses, highlighting key phrases or terms related to these themes.

• Researchers are instructed to summarize the main points for each theme, ensuring that each section includes a description of the theme along with relevant quotes from participants.

• Interaction with the chatbot (CHTP) is crucial; it helps identify specific topics and provides deeper analysis upon request.

Importance of Social Media Influence

• One significant theme identified is the influence of social media. CHTP offers an overview of perspectives and approximate percentages reflecting participant opinions on this topic.

• The analysis emphasizes both convergences (common agreements among respondents) and divergences (differing opinions), enriching the study's findings.

Convergences vs. Divergences

• It’s essential to not only focus on areas where most interviewees agree but also to explore differing viewpoints, which adds depth to the research.

• Clear descriptions of both convergences and divergences are necessary for comprehensive understanding.

Role of Direct Quotes in Contextualization

• Identifying direct quotes from participants is vital for contextualizing themes; for instance, one participant mentions using social media for information gathering.

• Researchers must contextualize these quotes within broader trends observed in previous studies regarding social media usage.

Participant Insights on Trust in News Sources

• Despite many stating they use social media for news, some participants express distrust towards information from these platforms, citing concerns over fake news.

• A notable quote highlights skepticism about Twitter as a reliable source due to prevalent misinformation.

Subtopics Related to Social Media

Subthemes Identified

• Two primary subthemes under "Role of Instagram and Twitter" include:

• Accessing quick and diverse information through these platforms.

• Concerns regarding trustworthiness and authenticity of content shared online.

Additional Themes Explored

• Other major themes discussed include:

• Social media as a digital agora (public space).

• Challenges associated with media literacy in today's digital landscape.

Advantages and Limitations of Using CHTP

Benefits Highlighted

• CHTP facilitates qualitative data exploration but does not replace human analytical depth or creativity.

Insights on AI and Human Creativity in Data Analysis

Limitations of AI in Thematic Analysis

• The initial data analysis by AI can identify thematic patterns, but it lacks the depth that human analysis provides. While AI can detect primary and secondary themes, it does not offer significant explanations.

• Current AI capabilities cannot replace human creativity; they provide a superficial understanding of data themes compared to human insight.

• Machines require detailed instructions for thematic analysis and cannot perform this task autonomously from the start, which consumes time and effort.

• There is a concern about the potential impact of AI on human creativity, as reliance on these tools may limit innovative thinking.

Discussion on Tools for Qualitative Data Analysis

• A question arises regarding the availability of teaching materials related to qualitative data analysis tools like ChatGPT and Atlas.ti.

• The speaker emphasizes that while Atlas.ti is often recommended for its simplicity, both tools have their advantages depending on user experience.

• It is noted that neither tool will solve issues with results; ultimately, humans must interpret findings regardless of the software used.

Comparing ChatGPT and Atlas.ti

• Atlas.ti offers various tags but lacks descriptive capabilities; ChatGPT provides better explanations and insights into data analysis processes.

• The speaker suggests trying ChatGPT due to its evolving models that promise improved performance in future iterations.

Future Considerations in Research Methodology

• As technology advances, there is an expectation that many tasks currently performed manually will eventually be automated by machines.

• A participant expresses interest in learning more about Atlas.ti but acknowledges the importance of considering other options like ChatGPT.

Addressing Concerns About AI Usage in Academia

• Participants discuss concerns regarding academic acceptance of AI tools. They highlight how traditional citation practices are similar to using AI assistance if properly acknowledged.

• Assurance is given that utilizing experienced guidance helps students navigate research challenges effectively when incorporating AI tools into their work.

Conclusion: Embracing Technology in Research

Discussion on AI in Research

Importance of Awareness in AI Utilization

• Emphasizes the need for vigilance regarding biases and errors when using artificial intelligence (AI) in research. The speaker notes that while researchers have the right to avoid AI, doing so may limit their access to valuable tools that enhance research quality.

Benefits and Risks of AI Tools

• Highlights that while there are potential misuses of AI, focusing on its positive applications is more beneficial. The speaker advocates for leveraging AI as a tool to facilitate better research outcomes rather than shunning it due to possible negative implications.

Compatibility of Thematic Analysis with Phenomenological Methodology

• A question arises about whether thematic analysis can coexist with phenomenological methodologies in thesis work. This indicates an interest in exploring diverse analytical frameworks within qualitative research.

Potential Applications of AI in Data Analysis

• Discusses the possibility of using AI for analyzing extensive data sets, such as summarizing lengthy transcripts or identifying key themes from large volumes of text. This suggests a practical application where AI can significantly ease the workload for researchers.

Closing Remarks and Future Classes