Uses and Risks of Artificial Intelligence in electoral campaigns 2023: Delphi survey of strategic experts in Colombia

Usos y Riesgos de la Inteligencia Artificial en las campañas electorales 2023: Encuesta Delphi a expertos estratégicos de Colombia

Usos e riscos da Inteligência Artificial nas campanhas eleitorais de 2023: inquérito Delphi a especialistas em estratégia na Colômbia

Marcela López-Ponce^1*
Daniel Barredo-Ibáñez^2**
Hada Sánchez Gonzales^3***

¹ Universidad de Sevilla, España.
² Universidad Fudan, Shanghái, China.
³ Universidad de Sevilla, España.
^* Digital Marketing Consultant. PhD from Universidad de Sevilla (US). Colombia. m@marcelalopez.co
^** Emergia Researcher, Universidad de Málaga (UMA), Spain. Researcher at Fudan University, Shanghai, China. Spain. daniel.barredo@uma.es
^*** Associate Professor of the Department of Journalism II at Universidad de Sevilla (US). Spain. misago@us.es

Abstract

Computational propaganda and bots have become tools within social networks for shaping opinions. However, the integration of artificial intelligence (AI) has added complexity to this landscape, allowing for natural language processing and the creation of automated strategies for political communication. While AI offers potential benefits, it also raises ethical concerns and the risk of spreading misinformation. Consequently, countries like Colombia are exploring the application of AI in political campaigns while simultaneously working on regulating its use within their democratic framework. In view of the 2023 upcoming elections, the debate surrounding the ethics and impact of AI in political communication gains significant relevance. Therefore, the aim of this study was to investigate the potential applications of artificial intelligence (AI) in Colombian political communication, its influence on polarization, and its role in disseminating disinformation. By employing the Delphi method with a panel of electoral strategy experts, the findings revealed a consensus among experts regarding AI's utility in creating content for social networks and segmenting political messages. However, there was disagreement regarding its suitability for generating government plans and its influence on the perception of the candidates' image. Additionally, experts concurred that algorithmic communication exacerbates polarization on social networks but held differing opinions on whether it enhances or diminishes electoral participation.

Keywords:
Artificial intelligence; political communication; ChatGPT; electoral processes; polarization; disinformation

Resumen

La propaganda computacional y los bots han surgido como herramientas en redes sociales para moldear opiniones. Sin embargo, la inteligencia artificial (IA) llega como actor para permitir el procesamiento del lenguaje natural y el desarrollo de estrategias de comunicación política automatizadas. Aunque promete beneficios, también plantea desafíos éticos y riesgos de desinformación. Países como Colombia exploran el uso de IA en campañas políticas, al tiempo que buscan regular su aplicación en su contexto democrático. Con la llegada de las elecciones 2023, el debate sobre la ética y el impacto de la IA en la comunicación política adquiere relevancia. Por ello, el objetivo de este estudio fue identificar los posibles usos de la IA en la comunicación política colombiana, su influencia en la polarización y su impacto en la desinformación. Se utilizó el método Delphi en un grupo de expertos en estrategia electoral. Los resultados arrojaron que concuerdan en que la IA es útil para crear contenido en redes sociales y segmentar mensajes políticos. No obstante, discrepan en considerar su uso para planes de gobierno y de su influencia en la percepción de la imagen de los candidatos. Además, coinciden en que la comunicación algorítmica aumenta la polarización en redes sociales, sin embargo, difieren en si fomenta o reduce la participación electoral.

Palabras clave:
Inteligencia artificial; comunicación política; ChatGPT; procesos electorales; polarización; desinformación

Resumo

A propaganda computacional e os bots surgiram como ferramentas nas redes sociais para moldar opiniões. No entanto, a chegada da inteligência artificial (IA) complexificou ainda mais o cenário, permitindo o processamento de linguagem natural e o desenvolvimento de estratégias de comunicação política automatizadas. Embora prometa benefícios, também coloca desafios éticos e riscos de desinformação. Assim, países como a Colômbia exploram o uso da IA em campanhas políticas, ao mesmo tempo em que buscam regulamentar sua aplicação em seu contexto democrático. Com as eleições de 2023 no horizonte, o debate sobre a ética e o impacto da IA na comunicação política ganha relevância. Portanto, o objetivo deste estudo foi explorar os possíveis usos da inteligência artificial (IA) na comunicação política colombiana, sua influência na polarização e seu impacto na desinformação. Assim, utilizando o método Delphi em um grupo de especialistas em estratégia eleitoral, os resultados mostraram que os especialistas concordam que a IA é útil para criar conteúdo em redes sociais e segmentar mensagens políticas. No entanto, discordam quanto ao seu uso para criar discursos ou planos de governo. Além disso, concordam que a comunicação algorítmica aumenta a polarização nas redes sociais, mas diferem quanto a se ela fomenta ou reduz a participação eleitoral.

Palavras-chave
Inteligência artificial; comunicação política; ChatGPT; processos eleitorais; polarização; desinformação

1. Introduction

Social and political polarization has become a tool for electoral strategists and a phenomenon of interest for researchers (Mora Ledesma, 2011). However, there are conflicting opinions about the effect of polarization on political participation. On one hand, some authors argue that it generates apathy and withdrawal (Fiorina & Abrams, 2008), while others assert that it stimulates the electorate and, by making positions explicit, inclines them toward participation (Lupu, 2015).

This is consistent with the observation that in recent decades, winning candidates and American voters have been moving in opposite directions, creating an interesting paradox of polarization (Galston, 2023). However, with the trend toward sensationalism, exacerbation, and dramatization, political polarization has led to the development of extreme political attitudes and positions (Lobo, 2017), as well as the creation of perverse incentives among power groups to favor conflict over cooperation with their opponents (Iyengar & Westwood, 2015). Nonetheless, some authors point out that research is still insufficient and limited regarding the relationship between polarization and democratic processes (Marino & Iannelli, 2023).

Consequently, with new possibilities, interest groups from various political currents have been inclined to develop different types of strategies on social networks, gaining relative influence in elections (Magnani, 2017). For example, during the 2016 United States presidential election, exposure to X reduced the percentage of Republican votes, with limited effects on participation and votes for the House of Representatives and the Senate (Fujiwara et al., 2020). On the other hand, in Taiwan, social networks had a positive impact on voter participation and attitude (Lee, 2020).

Additionally, computational propaganda emerges as another actor using automated algorithms to disseminate information shaped to serve interests or political agendas (Howard et al., 2023). Therefore, this research aims to identify the uses that political strategists make of these tools and their influence on polarization in Colombia.

1.1. Polarization and political communication

Social networks promote connectivity within groups; however, they also generate tendencies to ignore or attack information that contradicts partisan or ideological identity (Interian et al., 2023), consolidating identities and limiting boundaries between political groups (Iyengar & Westwood, 2015), sometimes even more accentuated than in the real world (Barredo-Ibáñez et al., 2021). This phenomenon, known as polarization, not only affects the dissemination and reach of political messages but is also driven by the ability of certain polarizing topics to influence opinions and connections in public discussions more than less divisive ones (Schuliaquer & Vommaro, 2020). This effect is intensified by messages that evoke feelings of anger and disgust, contributing to greater success in persuasion (Brady et al., 2019).

Exposure to partisan media increases ideological and emotional polarization through strategies like misinformation and hate speech (Vasist et al., 2023). Although partisan media is especially polarizing, the effect of exposure to non-partisan media is still unknown (Kubin & von Sikorski, 2021). In this sense, social networks promote hyper-emotionality in politics, hindering the creation of consensus between divergent parties (García, 2019). For example, Facebook and X function as virtual echo chambers, where individuals with similar opinions reinforce shared narratives that restrict political debate (Boxell et al., 2017; Cinelli et al., 2021). It has been suggested that multifactorial analysis models could be useful for recommending news and thus reducing polarization within these echo chambers (Treuillier et al., 2023).

1.2. Computational propaganda and political communication

The information age and cyberspace have created a new medium of communication that escaped media and government control (Castells, 2012; Roetzel, 2019). In this context, computational propaganda emerged as a strategic tool to create artificial currents of opinion (Bradshaw & Howard, 2017), later defined in academia as algorithmic communication (Campos-Domínguez & García-Orosa, 2018) and technopolitics (López-López & Oñate, 2019). The impacts associated with digital media strategies can vary depending on the political system. In emerging democracies, political participation and information consumption can be beneficial, while in established democracies, the decrease in political trust, increase in populism, and polarization can be detrimental (Lorenz-Spreen et al., 2023).

Computational propaganda is primarily conducted in electoral processes due to its high costs (Barredo-Ibáñez et al., 2021). Thanks to technological advances such as machine learning and sentiment analysis (Persily & Tucker, 2020), it has allowed for the creation of personalized content (Campos-Domínguez & García-Orosa, 2018). This has led to the automated generation of information by parties and candidates (Gutiérrez-Rubí, 2018), benefiting state propagandists and authoritarian tendencies (Trauthig et al., 2023).

1.3. Use of bots in political communication

Algorithmic communication, as a resource for electoral marketing strategies, can present significant risks to today's democratic society (López-López et al., 2023). First, algorithms can generate slander, leaving the media and citizens defenseless (Lewis et al., 2019), and second, their use has led to racist deviations through the dissemination of hate speech (Turner Lee, 2018).

At this point, the fabrication of consensus arises through the automated dissemination of propaganda via social bots (Persily & Tucker, 2020). These bots generate uncertainty and distortions by inflating reach metrics, promoting dominant narratives that foster negative emotions and challenges in network governance (Barredo-Ibáñez et al., 2021; Cai et al., 2022).

Cyber troops associated with government entities, political parties, or subcontracted services seek to influence public opinion in their favor (Bradshaw & Howard, 2017). For example, Politibot in Spain stands out for its informative quality (Sánchez & Sánchez, 2017; Sánchez & Sánchez, 2020), while in the United States, the influence of bots on users' political opinions has been observed (Bail et al., 2018).

In Ecuador, evidence of automated and human manipulation strategies on X was found during the 2017 presidential elections (Puyosa, 2017). During the 2020 presidential campaign, an increase in the conversion of commercial bots to electoral bots was observed, suggesting a greater use of this technology in political campaigns (Ackerman et al., 2020). In Mexico, "cyborgs" have been detected, which are bots capable of simulating human behavior (Clark et al., 2016).

During political events such as elections, the use of computational propaganda increases (Bradshaw et al., 2020). In Colombia, during the 2014 presidential elections, approximately 22% of users were bots (Cerón-Guzmán and León-Guzmán, 2015). In the 2018 presidential elections, cyber troops employing computational propaganda techniques were detected to influence public debate in favor of and against candidates (Bradshaw et al., 2021), although direct responsibility could not be attributed.

In Colombia, computational propaganda was detected in WhatsApp groups affiliated with Iván Duque, where distorted content was shared to improve his image and discredit other candidates (Chenou et al., 2020; Serrano, 2018; Worley, 2018).

1.4. Artificial intelligence, political communication, and misinformation

AI enables software to perform advanced tasks such as reasoning, learning, creating, and planning, comparable to humans (Google, n.d.; European Parliament, 2020). In politics, it is used in virtual assistants for speeches and campaigns (Dader & Campos Domínguez, 2017), allowing the automatic creation and dissemination of personalized messages by combining big data and advanced software (Kreiss & Jasinski, 2016).

AI is used to influence political and social opinion (Campos-Domínguez & García-Orosa, 2018; Mattelart & Vitalis, 2015). In Latin America, AI was used in electoral campaigns, such as in Quito, Ecuador, in 2019, where it helped Jaime Yunda win (Calderón, 2019). In fact, its role is evolving in governmental spheres: in Denmark, as the leader of a political party with Leader Lars (Jaimovich, 2023); and in Romania, as a social policy advisor to the government with Ion (Sánchez, 2023). Additionally, it is increasingly used to manipulate and generate political audiovisual content, such as the dissemination of a fake image of Donald Trump being arrested (Devlin & Cheetham, 2023) and a video from the Republican Party describing what the country would be like if Joe Biden were reelected as President of the United States (Seisdedos, 2023).

Moreover, AI has facilitated the creation of hyper-realistic audiovisual representations, known as deepfakes, which simulate the appearance and speech of real people (García-Ull, 2021). The first attempt to create a deepfake resulted from an application developed in 2017 by a Reddit user self-named “Deepfakes” (Cole, 2017) who used an autoencoder technique (Deshmukh & Wankhade, 2021). These techniques, combining deep learning and falsification, raise serious privacy concerns as they increase the risk of identity theft. Their use to manipulate faces poses a threat to politics (Pérez Dasilva et al., 2021), especially following the emergence of deepfakes of prominent figures like Barack Obama or Donald Trump (Devlin & Cheetham, 2023).

This scenario presents challenges in detecting fake news, as the growing sophistication of deepfakes makes it difficult to distinguish between real and fictitious content (Chawla, 2019).

1.5. Natural language model in political communication: ChatGPT

In this context, companies like Microsoft, Google, Meta, and OpenAI are competing to develop user-friendly platforms for natural language processing (Medina, 2023). Among them, ChatGPT from OpenAI has become popular, with over half of Americans having heard of it and 14% having used it (Vogels, 2023). ChatGPT is a model that generates coherent responses thanks to its ability to process large amounts of data (OpenAI, 2023).

Guadián (2023) suggests that the ChatGPT model has applications in areas such as marketing and politics, like drafting electoral speeches and government proposals. In India, its use is being explored in the legal sector through an AI committee in the Supreme Court (Gandhi & Talwar, 2023). On the other hand, lawmakers in Costa Rica used it to draft an AI regulation law (Sequeira, 2023). The debate about its regulation is gaining momentum in the United States and Europe (Bohórquez, 2023; Gonzalo, 2023).

1.6. Ethical challenges regarding the use of Artificial Intelligence

These advancements raise ethical debates and concerns about their potential manipulation (Laux, 2021). Although powerful, AI can also be more harmful than previous technologies in spreading fake news, as Corrales (2023) warns. There is concern about the impact of a deepfake's influence on a community and the public image of a leader.

In light of the new AI tools released to the general public, there have been calls to attention to the risks that widespread use may pose for electoral processes and how protection strategies need to be developed (Panditharatne & Giansiracusa, 2023). The ethical implications raise various concerns, ranging from privacy preservation, fairness, the need for accountability, to transparency in algorithmic discrimination processes (Big Data Marketer, 2023).

As philosopher John K. Roth warns, mere ethical discussion is not enough; we must act considering the context of an imperfect and unequal world (Hagerty & Rubinov, 2019). Therefore, according to the authors, AI systems will generate diverse and potentially global social impacts, with more marked consequences for vulnerable groups, necessitating the development of AI ethics that recognize these risks across different cultures.

In Colombia, although social networks do not seem to influence voting (Barredo Ibáñez et al., 2015), the regional elections of October 2023 could be the first instance of the massive use of these tools (Hoyos, 2023), affecting candidates' perception and generating credibility in fake news.

Given the above, this study aims to understand how experts perceive the challenges and resources related to the increasing use of computational propaganda, algorithmic communication, and artificial intelligence in electoral campaign strategies. The goal is to answer the following questions:

RG1: What are the possible uses of AI in political communication in Colombia?

RG2: To what extent does algorithmic communication influence the observed political polarization in Colombia?

2. Methodology

2.1. Research design

The study adopts a non-experimental cross-sectional approach with a descriptive scope, without exploratory objectives. It was conducted using the Delphi method, a technique used to establish guidelines and predict trends (López-Gómez, 2018). This method is characterized by its controlled feedback and statistical group consensus (Dalkey, 1972), allowing for the collection of group opinions on a topic through sequential questionnaires directed at experts (Simancas-González, 2016).

2.2. Sample description

The sampling profile was non-probabilistic by judgment (Galloway, 2005), which involved selecting sample members based on the knowledge and judgment of the researchers after consulting the database of the local mayor's office communication office. The selection of experts does not follow a common standard in terms of specific elements and the number of participants (López-Gómez, 2018; Steurer, 2011), as the number can vary depending on the research problem and available resources.

Four objective criteria were established to select digital political strategists in Colombia:

1. Degree in Social Communication or related fields.

2. At least 5 years of experience in managing social networks.

3. Experience in communication teams in electoral campaigns and/or during government tenure.

4. Being professionally active in political advisory roles.

Fifteen experts were identified (Table 1), of which 10 agreed to participate in the study, meeting the requirements described by López-Gómez (2018).

Given the specific characteristics of the Colombian context, it was deemed pertinent to include only professionals from this country. The group of experts maintains heterogeneity: men (N = 4) and women (N = 6), who have worked on political campaigns at the local level, in the Caribbean region (N = 5) and the Andean region (N = 5), as well as in national campaigns (N = 5). They have an average age of 34 years and 7 years of professional experience.

2.3. Data Collection Protocol

A questionnaire containing both closed and open-ended questions was administered. In the first round of the Delphi panel, 36 questions with a five-option scale were used, along with 6 open-ended questions to obtain more precise results (Table 2). The questionnaire was validated with the collaboration of two experts with PhDs in Communication and experience in political communication: Jennie Peña from the Universidad del Norte and Tania Cobos from the Universidad Tecnológica de Bolívar. Subsequently, a questionnaire was designed in Google Forms and distributed via email in July 2023.

In the second stage, a personalized questionnaire was designed for each expert, eliminating questions without 50% consensus in the first round and resending only those where responses differed from the majority. The results of the first questionnaire were provided to allow them to reconsider their responses, and an open-ended question was included at the end for any additional comments on their change or maintenance of decision.

In the quantitative analysis, considering definitive responses from the first and second rounds, those with 70% intragroup consensus or higher were selected, as this percentage is considered high according to the study by Romero-Collado (2021). The data were processed with SPSS, and variables were described using descriptive statistics such as frequency and percentage for greater precision (Steurer, 2011). Subsequently, for further analysis, Likert scale responses were grouped into three categories: "strongly agree” and “agree," "not sure," and "disagree” and “strongly disagree." Qualitative responses to open-ended questions were used to support quantitatively identified patterns.

3. Results

3.1. Uses of AI in Political Communication in Colombia

AI is used in political communication; however, the experts only reached a consensus on two items in this section. The respondents agree that natural language models are useful for creating content on candidates' social media and segmenting political messages (Table 3).

The experts indicated that AI helps to segment demographic data and characterize citizens, facilitating the generation of political speeches tailored to national interests. Additionally, they mentioned its use in Colombia since at least 2022 to create creative content on social media and develop effective personalized messages. They also agreed on its use in creating bots to spread negative messages, aligning with Lewis et al. (2019).

3.2. Influence of Algorithmic Political Communication on Political Polarization in Colombia

The obtained consensus highlights the influence of algorithmic political communication on polarization in Colombia. The experts agreed that the use of algorithmically generated strategies reinforces negative aspects such as the increase of polarization on social networks, the amplification of political divisions, the contribution to offline political polarization, and the promotion of conflict between opponents. From a positive perspective, the respondents agreed that these strategies can also promote the wider dissemination of positive messages (Table 4).

Four out of five scenarios of AI influence have negative connotations. Additionally, they assert that algorithmic political communication contributes to misinformation and the spread of fake news, polarizing citizens and favoring candidates or political parties instead of promoting democratic participation. It is also mentioned that AI could be used to detect and counteract the negative actions of opposing candidates.

3.3. Application of AI in Electoral Processes: Misinformation

Although the use of algorithmic political communication in electoral processes is associated with negative influence scenarios (Bender, 2022; Gagrčin et al., 2023), we found a total consensus of 100% among experts who believe it fosters voter participation. However, in contrast, 70% state that AI-generated content causes voter disinterest (see Table 5).

The experts pointed out that tools like chatbots and personalized messages can influence citizens when they go to vote. However, the rise of fake news by these same bots can generate disinterest, increase divisions among citizens, and promote hate speech that fuels the country's tensions.

Additionally, the experts see AI as useful in electoral processes: it detects misinformation, generates audiovisual content, and automates analysis. It also manipulates trends on X and opinions toward candidates (Table 6), as indicated by Lorenz-Spreen et al. (2023), with variable impacts depending on the political current.

The experts express concern about the potential of AI to spread misinformation and hate rather than strengthen democracy in Colombia. Although its utility in creating dynamic messages is recognized, the lack of ethical and legal regulation, including the use of software like ChatGPT, raises concerns.

On the other hand, there was no consensus on whether AI-generated content can structure a government plan or influence voters' perceptions of a candidate's image or the spread of hate speech (Table 7). These premises can be complemented with studies related to the influence and perception of candidates.

4. Discussions and Conclusions

The results reveal differing opinions on the role of AI in Colombian political communication. While its potential to enhance citizen participation is highlighted, concerns are also raised about its ability to restrict participation through the spread of hate speech or misinformation. There are worries that algorithmic strategies might increase polarization and conflict among opponents.

Contrary to Guadián's (2023) assertions, there was no consensus among the experts regarding whether natural language models like ChatGPT could be useful tools for creating campaign speeches and drafting government plans. Nevertheless, this phenomenon is already occurring in the country's political practice. These tools are improving their usage and results daily. For example, ChatGPT, through the use of prompts to obtain information, can be trained as a political consultant and perfect documents required for electoral campaign strategies.

Regarding the impact on democratic processes in Colombia, opinions diverge. While some studies suggest that polarization can reduce electoral participation (Fiorina & Abrams, 2008), others argue that it can promote it. However, Marino & Iannelli (2023) point out the uncertainty about the overall effects of polarization influenced by AI tools.

Given the above, it is relevant to consider the assertion that the use of AI tools in elections in Colombia may increase the spread of fake news, the manipulation of public opinion trends, and the discrediting of opposing candidates. This could encourage political groups to favor conflict over cooperation to achieve greater electoral influence, although the magnitude of its impact remains uncertain.

This information confirms that AI is a reality that directly influences the formation of opinions and citizens' decisions, as well as the quality of debate and democratic health. It is essential to consider how different political, cultural, and regulatory contexts affect the implementation and impacts of AI and to address the political consequences, especially misinformation and polarization. Additionally, the credibility that the use of AI can generate not only for a candidate but also for the political strategist, in terms of whether it is used as a complement to their work or performs the majority of their work for them, must be taken into account.

Finally, it is suggested to establish ethical boundaries and protective measures when using these tools in electoral campaigns in Colombia.

This study examines the impact of AI on politics and elections in Colombia, highlighting advantages and ethical concerns such as polarization and misinformation. It recommends investigating the relationship between AI and voting decisions, conducting quantitative studies on its influence on political polarization, and helping to establish ethical regulations for future elections. Additionally, comparing Colombia's situation with other countries regarding the use of AI in political communication and exploring how different contexts affect its implementation and effects would be useful.

The study has limitations in its sample and focus on expert opinions. Quantitative studies with representative samples of the population are recommended for more conclusive results. Future research could conduct controlled experiments and surveys during elections to inform ethical regulations in the use of AI.

Author’s Contribution

Marcela López Ponce: Investigation, formal analysis, methodology, data curation, validation, Writing-original draft. Daniel Barredo Ibáñez: Conceptualization, supervision, methodology and software, writing - review and editing. Hada Sánchez Gonzáles: Conceptualization, supervision, writing - review and editing. All authors have read and agree to the published version of the manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

References

Ackerman, John M., Cabo de la Vega, Antonio de, Casado, Fernando, Chávez V., Gonzalo, Hernández Zubizarreta, Juan, Luque, Ana, Macías, Javier, Maniglio, Franco, Mateos, Carlos, Merino, Antonio, Oliván, Fernando, Puglisi, Stefano, & Vera, Sergio (2020). La democracia de los trolls y el asalto a la democracia: Análisis del proceso electoral en Ecuador y los elementos irradiadores que lo conforman a través de sus redes sociales. Ediciones UTM. https://biblioteca-repositorio.clacso.edu.ar/handle/CLACSO/7917

Bail, Christopher A., Argyle, Lisa P., Brown, Taylor W., Bumpus, John P., Chen, Haohan, Hunzaker, M. B. Fallin, Lee, Jaemin, Mann, Marcus, Merhout, Friedolin, & Volfovsky, Alexander (2018). Exposure to opposing views on social media can increase political polarization. Proceedings of the National Academy of Sciences, 115(37), 9216-9221. https://doi.org/10.1073/pnas.1804840115

Barredo-Ibáñez, Daniel, De-la-Garza-Montemayor, Daniel-Javier, Torres-Toukoumidis, Ángel, & López-López, Paulo-Carlos (2021). Artificial intelligence, communication, and democracy in Latin America: A review of the cases of Colombia, Ecuador, and Mexico. Profesional de la Información, 30(6), Article e3006. https://doi.org/10.3145/epi.2021.nov.16

Barredo-Ibáñez, Daniel, Rivera González, José Gabriel, & Amézquita, Álex (2015). La influencia de las redes sociales en la intención de voto. Una encuesta a partir de las elecciones municipales en Ecuador 2014. Quórum Académico, 12(1), 136-154. https://dialnet.unirioja.es/servlet/articulo?codigo=5304980

Bender, Sarah (2022). Algorithmic Elections. Michigan Law Review, 121.3, 489. https://doi.org/10.36644/mlr.121.3.algorithmic

Big Data Marketer (2023, 15 de octubre). Los desafíos éticos en la aplicación de la inteligencia artificial: Un análisis profundo de las implicaciones actuales. [Entrada de Blog]. https://www.bigdata-social.com/desafios-de-la-etica-en-la-inteligencia-artificial-aplicada/

Bohórquez, Lucía (2023, mayo 16). “Si esta tecnología va mal, puede ir bastante mal”: Creador de ChatGPT pide regular IA. El Tiempo. https://www.eltiempo.com/tecnosfera/novedades-tecnologia/sam-altman-creador-de-chatgpt-pide-regulacion-de-ia-y-advierte-riesgos-769190

Boxell, Levi, Gentzkow, Matthew, & Shapiro, Jesse M. (2017). Is the internet causing political polarization? Evidence from demographics (Working Paper No. w23258). National Bureau of Economic Research. https://doi.org/10.3386/w23258

Bradshaw, Samantha, Bailey, Hannah, & Howard, Philip N. (2021). Industrialized Disinformation: 2020 Global Inventory of Organized Social Media Manipulation. Oxford Internet Institute. https://demtech.oii.ox.ac.uk/research/posts/industrialized-disinformation/

Bradshaw, Samantha, Campbell-Smith, Ualan, Henle, Amelie, Perini, Antonella, Shalev, Sivanne, Bailey, Hannah, & Howard, Philip N. (2020). Country Case Studies Industrialized Disinformation: 2020 Global Inventory of Organized Social Media Manipulation. https://demtech.oii.ox.ac.uk/wp-content/uploads/sites/12/2021/03/Case-Studies_FINAL.pdf

Bradshaw, Samantha, & Howard, Philip N. (2017). Troops, Trolls and Troublemakers: A Global Inventory of Organized Social Media Manipulation. Computational Propaganda Research Project. https://ora.ox.ac.uk/objects/uuid:cef7e8d9-27bf-4ea5-9fd6-855209b3e1f6

Brady, William J., Wills, Julian A., Burkart, Dominic, Jost, John T., & Van Bavel, Jay J. (2019). An ideological asymmetry in the diffusion of moralized content on social media among political leaders. Journal of Experimental Psychology: General, 148, 1802-1813. https://doi.org/10.1037/xge0000532

Cai, Meng, Luo, Han, Meng, Xiao, & Cui, Ying (2022). Differences in Behavioral Characteristics and Diffusion Mechanisms: A Comparative Analysis Based on Social Bots and Human Users. Frontiers in Physics, 10, 875574. https://www.frontiersin.org/articles/10.3389/fphy.2022.875574

Calderón, María José (2019). The use of social media and AI to shape democracy in Ecuador. En Alan Finlay Global Information Society Watch (pp. 124-128). Association for Progressive Communications. https://giswatch.org/node/6169

Campos-Domínguez, Eva, & García-Orosa, Berta (2018). Comunicación algorítmica en los partidos políticos: Automatización de producción y circulación de mensajes. El Profesional de la Información, 27(4), 769. https://doi.org/10.3145/epi.2018.jul.06

Castells, Manuel (2012). Redes de indignación y esperanza: Los movimientos sociales en la era de Internet. Alianza.

Cerón-Guzmán, Jhon Adrián, & León, Elizabeth (2015). Detecting Social Spammers in Colombia 2014 Presidential Election. En O. Pichardo Lagunas, O. Herrera Alcántara, & G. Arroyo Figueroa (Eds.), Advances in Artificial Intelligence and Its Applications (pp. 121-141). Springer International Publishing. https://doi.org/10.1007/978-3-319-27101-9_9

Chawla, Ronit (2019). Deepfakes: How a pervert shook the world. International Journal of Advance Research and Development, 4(6), 4-8.

Chenou, Jean-Marie, Cabarcas Velandia, Daniel, & Sepulveda, Maria Nicoll (2020). Social media and political polarization in Latin America: Analyzing online discussions during the 2018 presidential campaign in Colombia 1. En The Politics of Technology in Latin America (Volume 2). Routledge. https://doi.org/10.4324/9780429343247-11

Cinelli, Matteo, De Francisci Morales, Gianmarco, Galeazzi, Alessandro, Quattrociocchi, Walter, & Starnini, Michele (2021). The echo chamber effect on social media. Proceedings of the National Academy of Sciences, 118(9), e2023301118. https://doi.org/10.1073/pnas.2023301118

Clark, Emily M., Williams, Jake R., Jones, Cameron A., Galbraith, Robert A., Danforth, Christopher M., & Dodds, Peter Sheridan (2016). Sifting robotic from organic text: A natural language approach for detecting automation on Twitter. Journal of Computational Science, 16, 1-7. https://doi.org/10.1016/j.jocs.2015.11.002

Cole, Samantha (2017, 11 de diciembre). AI-Assisted Fake Porn Is Here and We’re All Fucked. Vice. https://motherboard.vice.com/en_us/article/gydydm/gal-gadot-fake-ai-porn accessed 10 September 2021.

Corrales, Diego (2023). Campañas ganadoras. Bronce.

Dader, José Luis, & Campos Domínguez, Eva (Eds.) (2017). La búsqueda digital del voto: Cibercampañas electorales en España 2015-16. Tirant lo Blanch. https://dialnet.unirioja.es/servlet/libro?codigo=702481

Dalkey, Norman Crolee (1972). Studies in the quality of life: Delphi and decision-making. Lexington Books.

Deshmukh, A., & Wankhade, S. B. (2021). Deepfake detection approaches using deep learning: A systematic review. En V. E. Balas, V. B. Semwal, A. Khandare, & M. Patil (Eds.), Intelligent computing and networking (pp. 285-295). Springer. https://doi.org/10.1007/978-981-15-7421-4_27

Devlin, Kayleen, & Cheetham, Joshua (2023, marzo 25). Cómo detectar imágenes creadas por inteligencia artificial como las fotos falsas del arresto de Trump. BBC News Mundo. https://www.bbc.com/mundo/noticias-65071726

European Parliament (2020, September 8). ¿Qué es la inteligencia artificial y cómo se usa? https://www.europarl.europa.eu/news/es/headlines/society/20200827STO85804/que-es-la-inteligencia-artificial-y-como-se-usa

Fiorina, Morris P., & Abrams, Samuel J. (2008). Political Polarization in the American Public. Annual Review of Political Science, 11(1), 563-588. https://doi.org/10.1146/annurev.polisci.11.053106.153836

Fujiwara, Thomas, Müller, Karsten, & Schwarz, Carlo (2020). The Effect of Social Media on Elections: Evidence from the United States. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3719998

Gagrčin, Emilija, Ohme, Jakob, Buttgereit, Lena, & Grünewald, Franziska (2023). Datafication Markers: Curation and User Network Effects on Mobilization and Polarization During Elections. Media and Communication, 11(3). https://doi.org/10.17645/mac.v11i3.6641

Galloway, Alison (2005). Non-Probability Sampling. En Encyclopedia of Social Measurement (pp. 859-864). Elsevier. https://doi.org/10.1016/B0-12-369398-5/00382-0

Galston, William (2023, january 20). The polarization paradox: Elected officials and voters have shifted in opposite directions. Brookings. https://www.brookings.edu/articles/the-polarization-paradox-elected-officials-and-voters-have-shifted-in-opposite-directions/

Gandhi, Parth, & Talwar, Vikram (2023). Artificial intelligence and ChatGPT in the legal context. Indian Journal of Medical Sciences, 75, 1-2. https://doi.org/10.25259/IJMS_34_2023

García, Javier (2019, mayo 14). Polarización política: El fenómeno que debería estar en boca de todos (434; Dossier, pp. 31-33). CaixaBank Research. https://www.caixabankresearch.com/es/economia-y-mercados/sector-publico/polarizacion-politica-fenomeno-deberia-estar-boca-todos

García-Ull, F. J. (2021). Deepfakes: el próximo reto en la detección de noticias falsas. Anàlisi: Quaderns de Comunicació i Cultura, 64, 103-120. https://doi.org/10.5565/rev/analisi.3378

Gonzalo, María (2023, abril 27). La Unión Europea prepara la normativa de inteligencia artificial que regulará a ChatGPT. Newtral. https://www.newtral.es/ley-inteligencia-artificial-ia-chatgpt-union-europea-ue/20230427/

Google (s/f). ¿Qué es la inteligencia artificial o IA? | Google Cloud. Google Cloud. https://cloud.google.com/learn/what-is-artificial-intelligence?hl=es-419

Guadián, Carlos (2023). Del spin doctor al spin prompter. ACOP, 81, 5-8. https://compolitica.com/n81-et-2-chatgpt-y-comunicacion-politica-una-ia-para-controlar-el-mensaje/

Gutiérrez-Rubí, Antoni (2018, febrero 26). ¿Asistentes virtuales también para partidos y políticos? El País. https://elpais.com/politica/2018/02/26/micropolitica/1519636154_945042.html

Hagerty, Alexa, & Rubinov, Igor (2019). A Review of the Social Impacts and Ethical Implications of Artificial Intelligence. Global AI Ethics. https://arxiv.org/abs/1907.07892

Howard, Philip, Lin, Frank, & Tuzov, Valery (2023). Computational propaganda: Concepts, methods, and challenges. Communication and the Public, 8(2), 47-53. https://doi.org/10.1177/20570473231185996

Hoyos, Federico (2023, abril 11). Inteligencia artificial: ¿Aliada o enemiga de la democracia? www.elcolombiano.com. https://www.elcolombiano.com/opinion/columnistas/inteligencia-artificial-aliada-o-enemiga-de-la-democracia-FF21066344

Interian, Roberto, G. Marzo, Roberto, Mendoza, Ignacio, & Ribeiro, César C. (2023). Network polarization, filter bubbles, and echo chambers: An annotated review of measures and reduction methods. International Transactions in Operational Research, 30(6), 3122-3158. https://doi.org/10.1111/itor.13224

Iyengar, Shanto, & Westwood, Sean J. (2015). Fear and Loathing across Party Lines: New Evidence on Group Polarization: FEAR AND LOATHING ACROSS PARTY LINES. American Journal of Political Science, 59(3), 690-707. https://doi.org/10.1111/ajps.12152

Jaimovich, Ariel (2023, marzo 21). Una charla con el robot que lidera el Partido Sintético en Dinamarca y que busca un escaño en el parlamento. La Nación. https://www.lanacion.com.ar/tecnologia/una-charla-con-leader-lars-el-robot-que-lidera-el-partido-sintetico-en-dinamarca-y-que-busca-darle-nid21032023/

Kreiss, Daniel, & Jasinski, Christine (2016). The Tech Industry Meets Presidential Politics: Explaining the Democratic Party’s Technological Advantage in Electoral Campaigning, 2004-2012. Political Communication, 33(4), 544-562. https://doi.org/10.1080/10584609.2015.1121941

Kubin, Emily, & von Sikorski, Christian (2021). The role of (social) media in political polarization: A systematic review. Annals of the International Communication Association, 45(3), 188-206. https://doi.org/10.1080/23808985.2021.1976070

Laux, Johannes (2021). OII | Should we worry about the government nudging us with personalised advertising? Oxford Internet Institute. https://www.oii.ox.ac.uk/news-events/news/should-we-worry-about-the-government-nudging-us-with-personalised-advertising

Lee, Hyunjin (2020). Voters’ involvement, attitude, and confidence in the era of new media. Palgrave Communications, 6(1), Article 1. https://doi.org/10.1057/s41599-019-0368-9

Lewis, Seth C., Guzman, Andrea L., & Schmidt, Thomas R. (2019). Automation, Journalism, and Human–Machine Communication: Rethinking Roles and Relationships of Humans and Machines in News. Digital Journalism, 7(4), 409-427. https://doi.org/10.1080/21670811.2019.1577147

Lobo, Sebastián (2017). Cómo influyen las redes sociales en las elecciones. Nueva Sociedad | Democracia y política en América Latina, 269. https://nuso.org/articulo/como-influyen-las-redes-sociales-en-las-elecciones/

López-Gómez, Ernesto (2018). El método Delphi en la investigación actual en educación: una revisión teórica y metodológica. Educación XX1, 21(1). https://revistas.uned.es/index.php/educacionXX1/article/view/20169

López-López, Paulo-Carlos, Barredo-Ibáñez, Daniel, & Jaráiz-Gulías, Erika (2023). Research on Digital Political Communication: Electoral Campaigns, Disinformation, and Artificial Intelligence. Societies, 13(5), 126. https://doi.org/10.3390/soc13050126

López-López, Paulo-Carlos, & Oñate, Pablo (2019). De la videopolítica a la ciberpolítica: Debate entre candidatos y televisiones en cinco elecciones presidenciales. El Profesional de la Información, 28(5), e280512. https://doi.org/10.3145/epi.2019.sep.12

Lorenz-Spreen, Philipp, Oswald, Lisa, Lewandowsky, Stephan, & Hertwig, Ralph (2023). A systematic review of worldwide causal and correlational evidence on digital media and democracy. Nature Human Behaviour, 7(1), Article 1. https://doi.org/10.1038/s41562-022-01460-1

Lupu, Noam (2015). Party Polarization and Mass Partisanship: A Comparative Perspective. Political Behavior, 37(2), 331-356. https://doi.org/10.1007/s11109-014-9279-z

Magnani, Esteban (2017). Big data y política. El poder de los algoritmos. Nueva Sociedad, 269. https://nuso.org/articulo/big-data-y-politica/

Marino, Giada, & Iannelli, Laura (2023). Seven years of studying the associations between political polarization and problematic information: A literature review. Frontiers in Sociology, 8, 1174161. https://doi.org/10.3389/fsoc.2023.1174161

Mattelart, Armand, & Vitalis, André (2015). De Orwell al cibercontrol. Gedisa.

Medina, Edgar (2023, mayo 25). Las 10 empresas líderes en inteligencia artificial en 2023 [CrónicaTech]. https://cronica.tech/tecnologia/software/las-empresas-lideres-en-inteligencia-artificial-y-machine-learning-en-2023/

Mora Ledesma, Martín (2011). Metodología de la comunicación política en campañas electorales (1. ed). Plaza y Valdés Editores.

OpenAI (2023, february 1). Introducing ChatGPT Plus. OpenAI. https://openai.com/blog/chatgpt-plus

Panditharatne, Manar, & Giansiracusa, Noah (2023, julio 21). Cómo la inteligencia artificial pone en riesgo las elecciones y las medidas que se requiren para protegernos | Brennan Center for Justice. Brennan Center. https://www.brennancenter.org/es/our-work/analysis-opinion/inteligencia-artificial-pone-en-riesgo-elecciones-medidas-proteger-democracia

Pérez Dasilva, Jesús Ángel, Meso Ayerdi, Koldobika, & Mendiguren Galdospin, Terese (2021). Deepfakes on Twitter: Which Actors Control Their Spread? Media and Communication, 9(1), 301-312. https://doi.org/10.17645/mac.v9i1.3433

Persily, Nathaniel, & Tucker, Joshua A. (Eds.) (2020). Social Media and Democracy: The State of the Field, Prospects for Reform (1a ed.). Cambridge University Press. https://doi.org/10.1017/9781108890960

Puyosa, Iria (2017). Bots políticos en Twitter en la campaña presidencial #Ecuador2017. Contratexto, 27(027), 39-60. https://doi.org/10.26439/contratexto.2017.027.002

Roetzel, Peter Gordon (2019). Information overload in the information age: A review of the literature from business administration, business psychology, and related disciplines with a bibliometric approach and framework development. Business Research, 12(2), 479-522. https://doi.org/10.1007/s40685-018-0069-z

Romero-Collado Ángel (2021). Essential elements to elaborate a study with the (e)Delphi method. Enferm Intensiva (Engl Ed), 32(2), 100-104. https://doi.org/10.1016/j.enfi.2020.09.001

Sánchez, Raúl (2023, marzo 27). Rumania designa a una inteligencia artificial como consejera del Gobierno. El País. https://elpais.com/sociedad/2023-03-27/rumania-designa-a-una-inteligencia-artificial-como-consejera-del-gobierno.html

Sánchez-Gonzales, Hada M., & Sánchez-González, María (2017). Los bots como servicio de noticias y de conectividad emocional con las audiencias. El caso de Politibot. Doxa Comunicación. Revista interdisciplinar de estudios de comunicación y ciencias sociales, 63-84. https://doi.org/10.31921/doxacom.n25a3

Sánchez-Gonzales, Hada M., & Sánchez-González, María (2020). Bots conversacional en la información política desde la experiencia de los usuarios: Politibot. Communication & Society, 33(4), Article 4. https://doi.org/10.15581/003.33.4.155-168

Schuliaquer, Iván, & Vommaro, Gabriel (2020). Introducción: La polarización política, los medios y las redes. Coordenadas de una agenda en construcción. REVISTA SAAP, 14(2). https://doi.org/10.46468/rsaap.14.2.I

Seisdedos, Iker (2023, abril 26). Los republicanos atacan a Biden con un vídeo de inteligencia artificial que imagina qué pasaría si gana. El País. https://elpais.com/internacional/2023-04-26/los-republicanos-atacan-a-biden-con-un-video-de-inteligencia-artificial-que-imagina-que-pasaria-si-gana.html

Sequeira, Aarón (2023, mayo 30). 5 diputados utilizan Chat GPT4 para construir proyecto de ley sobre inteligencia artificial. La Nación. https://www.nacion.com/el-pais/politica/5-diputados-utilizan-chat-gtp4-para-presentar/2RECU7GDTBHXPI3M4LPJBNOI2Y/story/

Serrano, Sebastián (2018, marzo 5). La máquina de noticias engañosas que ayuda a Iván Duque. Cerosetenta. https://cerosetenta.uniandes.edu.co/la-maquina-de-noticias-enganosas-que-ayuda-a-ivan-duque/

Simancas-González, Esther (2016). La comunicación en la universidad pública española. De la comunicación instrumental a la comunicación para la democracia y la participación [Universidad de Málaga]. https://riuma.uma.es/xmlui/handle/10630/12246

Steurer, Johann (2011). The Delphi method: An efficient procedure to generate knowledge. Skeletal Radiology, 40(8), 959-961. https://doi.org/10.1007/s00256-011-1145-z

Trauthig, Inga Kristina, Martin, Zelly Celeste, & Woolley, Samuel Clark (2023). Messaging Apps: A Rising Tool for Informational Autocrats. Political Research Quarterly, 10659129231190932. https://doi.org/10.1177/10659129231190932

Treuillier, Célina, Brun, Armelle, & Castagnos, Sylvain (2023). A Multi-factorial Analysis of Polarization on Social Media. Adjunct Proceedings of the 31st ACM Conference on User Modeling, Adaptation and Personalization, 1-6. https://doi.org/10.1145/3563359.3597393

Turner Lee, Nicol (2018). Detecting racial bias in algorithms and machine learning. Journal of Information, Communication and Ethics in Society, 16(3), 252-260. https://doi.org/10.1108/JICES-06-2018-0056

Vasist, Pramukh Nanjundaswamy, Chatterjee, Debashis, & Krishnan, Satish (2023). The Polarizing Impact of Political Disinformation and Hate Speech: A Cross-country Configural Narrative. Information Systems Frontiers. https://doi.org/10.1007/s10796-023-10390-w

Vogels, Emily A. (2023). A majority of Americans have heard of ChatGPT, but few have tried it themselves. Pew Research Center. https://www.pewresearch.org/short-reads/2023/05/24/a-majority-of-americans-have-heard-of-chatgpt-but-few-have-tried-it-themselves/

Worley, Will (2018, june 13). Misinformation Is Shaping the Colombian Election. Centre for International Governance Innovation. https://www.cigionline.org/articles/misinformation-shaping-colombian-election/