Abstract
With the disruptive emergence of Artificial Intelligences in the production of knowledge, new questions arise about the role that these technologies should occupy within the different spheres through which the past is represented. This article examines ChatGPT’s potential as a pedagogical tool for the development of historical thinking. For this purpose, we present the analysis of an interactive process around the historical event usually referred to as the Spanish Reconquista (711–1492). Through argumentative interventions, a dialogue is constructed in which the information provided by the AI is questioned, proving its ability to rectify and incorporate elements of historical thinking into its discourse. An ethical and multidisciplinary perspective is needed, which safeguards the responsible and beneficial use of these technologies in the construction and transmission of historical knowledge.
The intersection between generative Artificial Intelligences (AI) and making sense of the past has given rise to fertile ground for reflections on how digital advances shape the way we understand, transmit and teach history in the digital age. From the earliest cave paintings to contemporary museums, humans have occupied a unique role in the preservation and transmission of our past: our cognitive abilities, or what we can call ‘natural intelligence’, grant us as a species the unique ability to represent historical events through different formats. However, recent advances in AI challenge this exclusivity.
The emergence of these tools poses a disruption to the way we create and transmit knowledge: programs such as ChatSonic® or Google Bard® have the ability to perform complex linguistic tasks surpassing analogue human productions in speed and quality. In the audiovisual field, tools such as DALL-E® or Runaway® are capable of producing artistic products out of textual descriptions, challenging traditional conventions about creative processes. These technologies are not only altering the way in which linguistic and audiovisual content is produced but are becoming co-agents of originally human productions.
‘What we are talking about is potentially the end of human history. Not the end of history, just the end of its human-dominated part’ (Harari, 2023, para. 10). The emergence of technological tools capable of learning, developing intelligence and, consequently, demonstrating agency in decision-making opens the door to an ethical questioning about the role of AIs in our societies, as well as the use and control we should have over them.
Governments, companies, schools, universities and hospitals are faced with the decision to incorporate or censor the use of this tool, generating a multitude of debates about the opportunities and risks associated with its incorporation into daily activities. Experts from different disciplines, through the Future of Life Institute (2023), have positioned themselves in this debate by calling for an immediate pause in the training of AI systems superior to GPT-4 (Generative Pre-Trained Transformer).
GPT-4, commercially known as ChatGPT®, has been one of the most disruptive AI tools, breaking the record for the most rapidly adopted technology in human history (Brascia, 2023). Among its multitude of applications, this AI has the ability to tackle tasks such as writing scientific papers, composing poems, solving mathematical formulas, developing business plans and generating computer programming code. Its versatility in understanding and generating natural language has made it a powerful tool in many sectors, transforming the way we interact with information and technology.
This AI is based on a language model that has been trained with a vast amount of text to generate coherent and relevant responses based on the information provided by the user. Given the translation it performs from our written language to computational language and vice versa, this tool falls within the scope of Natural Language Processing (NLP). Its revolutionary workings usher in an active and generative technology that interacts with the user, adapting its responses to highly specific and complex conversations and contexts.
Understanding the impact of generative AI on the preservation, transmission and generation of historical knowledge requires a comprehensive view of the historical culture of society (Carretero et al., 2017). Thus, we can distinguish three main spheres of representation of the past: the academic sphere, guarded by historiographic research; the popular sphere, shaped by the generation and transmission of cultural products (monuments, traditions, films, video games. . .); and the educational sphere, represented by teaching-learning processes in formal and informal contexts (schools, museums and heritage sites). These three pillars interact dynamically and influence each other, shaping the collective understanding of the past.
From this perspective, the teaching of history is presented as a field of study that includes, influences and is influenced by the rest of the spheres of representation of the past. It follows that the objective of history education is not to generate future historians but to offer the student an interpretative framework of the events, causes, consequences, concepts and social structures that have shaped the foundations of our social reality. Historical thinking is presented in this sense as a necessary skill for the critical analysis of the past, as well as for the future application of the cultural baggage contained in the teaching-learning processes. From this perspective, the central objective of history teaching would not be for students to store endless data but for them to be able to reflect on the past in a nuanced way (Carretero, 2024; Seixas, 2017).
What kind of implications might have the incorporations of IAs on the field of knowledge transmission? What ethical challenges does this new era of digital transformation bring, where AIs may become co-agents of our cultural productions? In this article we will analyse the implications of ChatGPT® in the transmission and teaching of history, reflecting on the introduction of this tool in the different spheres of representation of the past.
The aim of this article is to explore the possibilities and limitations of the ChatGPT® tool in the teaching of history, taking into account its natural language processing capabilities. To this end, we present an analysis of the interaction carried out with the chatbot about the Spanish Reconquista, a historical event which illustrates the different narratives that influence the understanding of historical knowledge. In this context, the potential of this AI to generate spaces for dialogue that facilitate the development of historical thinking in students is addressed. It is important to note that often national narratives, termed master narratives or foundational narratives (Carretero & van Alphen, 2017; Wertsch, 2002), more closely resemble myths of origin, considered a basic symbolic support of the political construction of nations by historians (Hobsbawm & Ranger, 1983), than historiographical explanations proper. Therefore, from the point of view of teaching historical thought, a paramount objective is teaching how to deconstruct it.
Finally, some of the ethical challenges posed by the introduction of this tool in the different fields of history will be discussed, proposing a series of conclusions for a critical approach to this new era of technological revolution.
ChatGPT®, teaching history and historical thinking
In the field of history teaching, the debates about what to teach (the contents) and how to teach it (the pedagogical tools) have been modifying the school curriculum since the establishment of the first educational systems. This has involved a series of fluctuations in the ways of representing the past and, consequently, variations in the contents transmitted and the methodologies applied.
Given the research carried out, we can differentiate two objectives underlying these fluctuations (Carretero, 2011, 2021): the romantic objective (stemming from Romanticism) and the enlightened objective (stemming from the Enlightenment current of thought). From the educational point of view, the pursuit of these objectives has substantially influenced citizens’ approach to historical knowledge, conditioning their representations and interpretations of past events.
The romantic vision of history teaching emerged at the end of the nineteenth century, when modern nations began to design and implement their educational systems. With the strategic aim of consolidating and strengthening the image of the nation (Anderson, 1991), historical narratives were transmitted to instill in students a sense of belonging and patriotic fervor (Nussbaum & Cohen, 2002). This entailed reducing and simplifying content based on a single master narrative (Barton & Levstik, 2004; VanSledright, 2008), and the consequent suppression of alternative narratives that could jeopardize it. The official history of the nation was thus transmitted as an unquestionable reality, immune to the passage of time.
In the mid-twentieth century, the concept of national identity came under scrutiny by the field of historiographic study (Álvarez-Junco, 2011; Gellner, 1978). The teaching of history ceased to be conceived as a political instrument for the benefit of the nation, stating that its objective should lie in the development of students’ critical thinking. This paradigm shift entailed the incorporation of academic and cognitive aspects of history to the educational sphere, introducing historians’ own methodologies in their work of analysing the past (Wineburg, 2001). Thus, the enlightened perspective of history teaching seeks to train citizens who are critical of the past, capable of participating effectively in the historical changes taking place at the national and international levels.
Thinking like a historian, or thinking historically, implies a series of skills and approaches that go beyond the simple memorization of historical facts, such as dates, names or concrete facts. This type of knowledge is called first-order concepts because of its reference to the basic elements that make up the historical narrative; without these, it would be difficult to contextualize and understand the space-time in which past events are located.
However, a deep and critical understanding of history requires, in addition, the ability to analyse primary and secondary sources critically, assess the reliability of information, understand historical context and construct coherent historical narratives based on available evidence. It involves questioning assumptions, exploring multiple perspectives and recognizing the complexity of historical events and processes (Freedman, 2015; Lévesque, 2008; Nordgren, 2016; Seixas, 2017; Wineburg, 2001). In other words, thinking historically demands that the student be able to identify the existence of second-order concepts.
There is considerable agreement on the skills linked to these second-order concepts (Carretero & Pérez-Manjarrez, 2022; Lévesque & Clark, 2018; Retz, 2017). Thus, thinking historically requires:
(a) Being able to use evidence to confirm or disprove single facts about the past.
(b) Understanding that historical accounts are not copies of single facts about the past but constructions that are subsequently elaborated.
(c) Imagining past situations that cannot be experienced in the present, understanding that beliefs about the world which are not currently shared can, however, exist.
(d) Developing hypotheses about the multi-causality of past events.
(e) Defining abstract concepts precisely and understanding the fluctuations of meanings of these concepts as a function of time.
(f) Examining the extent to which the developed hypotheses fit the facts.
(g) Analysing change and continuity, understanding ideas related to time such as duration, sequence and temporal conventions.
For the development of reflective and democratic societies, history in the classroom must do more than simply teach first-order concepts; it must also teach students to think historically, understanding second-order concepts (Carretero & Pérez-Manjarrez, 2022). That is, the teaching of history should orient its pedagogical task not only towards the learning of concrete data but to the development of ways of thinking and reasoning as actually carried out by professional historians (Carretero, 2017), without, however, striving to turn students into little historians, as was said above. In this sense, these objectives would be valid not just for students but also for citizens in general, since historical representations also play an important role in the construction of citizenship.
Reflecting on the use of ChatGPT® in history teaching therefore implies asking: Could this AI serve as an educational tool for the development of historical thinking? Could AI facilitate or hinder historical thinking skills?
Several approaches have emerged towards the study of this issue. Carrasco Rodríguez (2023) has explored the possible contributions of ChatGPT® in the process of creating educational resources and implementing innovative didactic approaches. After using AI for the generation of theoretical content, exercises and practical activities around the teaching of the history of Charles V, this author recognized the usefulness of this AI for easing teaching work, adapting content and bringing forth pedagogical innovation in view of a greater incorporation of historical thinking in the classroom.
Tirado-Olivares et al. (2023), on the other hand, analysed the argumentative skill of ChatGPT® in written essays, taking into account the elements of historical thinking defined by Seixas and Morton (2013). To do so, they asked university students and future history teachers to evaluate two essays without knowledge of their authorship: the first one prepared by a university student and the second one generated by ChatGPT®. The results show that the subjects judged the AI-generated essay as better in terms of historical thinking and application of second-order concepts.
Both examples, as well as similar ones (Lyell, 2023; Nguyen et al., 2023), focus their analysis on the quality of the responses offered by ChatGPT®, showing that its natural processing capacity is similar to and adapted to existing pedagogical actions. However, beyond knowing what learning products the chatbot can generate, it is necessary to ask what learning processes can facilitate the development of historical thinking for the students interacting with the tool.
The quality that distinguishes ChatGPT® from other types of technology is its ability to process, understand and produce natural language, generating interactive spaces that transcend the mere reception and reproduction of knowledge. This enables the construction of dialogues between the AI and the user, taking into account not only the words but also the context, intentionality and fluency, all of them characteristics of human conversations. Dialogue is presented not only as a means of transmitting information but also as a fundamental way to construct meanings and promote deeper and more reflective learning.
In the realm of history education, dialogic processes can promote historical thinking to the extent that they encourage the exchange of perspectives, debate and discussion. This allows students to question assumptions, evaluate multiple sources and perspectives and recognize the complexity of events (Carretero & Pérez-Manjarrez, 2024; van Boxtel & van Drie, 2017). Given the functional nature of ChatGPT®, could its use generate spaces for dialogue that foster historical thinking?
To explore this possibility, we have selected an event that clearly portrays the confrontation between the romantic objective and the enlightened objective of history teaching: the Spanish Reconquista.
The period from 718 to 1492 is known as the Reconquista, a period in which a series of battles took place in the Iberian Peninsula between the Christian kingdoms and the Islamic kingdoms. In the year 711, the Muslims took power in the peninsula after defeating the Visigothic kingdom, then dominant in the territory. As the Reconquista has traditionally been presented, both by historiography itself and by the historical culture in general, this led eight years later to the beginning of a process of ‘recovery of the territory’ by the Christian kingdoms that lasted, according to this vision, almost 800 years. Despite the fact that during this period the nation of Spain did not yet exist, the historiographic interpretations of the following centuries led this narrative to become a historical reference of identity and foundational character. Thus, the Spanish national identity of the nineteenth century found in the Reconquista a narrative about the possession of territories and the capacity of Spaniards to ‘recover what was theirs’.
Recently, the work of Ríos Saloma (2005), among others, has been able to demonstrate that the term Reconquista does not appear in Spanish historiography until the eighteenth century, three centuries after the end of the event. Those who had previously been referred to as ‘Visigoths’, ‘Asturians’ or ‘Cantabrians’ in historical texts became only ‘Spaniards’, thus legitimizing a struggle against the Muslims to ‘recover Spanish territory’ at a time when the nation of Spain did not yet exist. Table 1 presents the conceptual dimensions derived from the use of this term, in comparison with the evidence provided by historiographical research.
Comparison between the conceptual dimensions of the romantic vision and the enlightened vision of the Spanish Reconquista. Extracted from Lopez et al. (2015).
Although this master narrative, of a romantic nature, is no longer present in Spanish history textbooks, the term Reconquista continues to be used frequently in a multitude of areas of cultural representation (museums, political discourse, media and video games, among others). Thus, there is a coexistence between the enlightened objectives of the teaching of contemporary history and the romantic conceptions of this event.
Interviews conducted with young students of Spanish origin reveal how their oral narratives about the Reconquista are still predominantly sustained by romantic arguments from the past (Lopez et al., 2015).
The Arabs invade a territory that is not theirs. For more than seven centuries they keep trying to conquer what is all Spanish territory. And the Spaniards. . . reconquered it again to make it once more theirs when in fact it was, in essence, their territory before the Arabs entered. (Juan, 25 years old)
Of the 31 interviews conducted, 80.6% interpreted the Reconquista event as a process of loss and recovery of a pre-existing national territory. Spanish nationality and the possession of peninsular territory are presented as constant variables over time, reinforcing the strength of these historical arguments.
[Why do you think the Christians wanted to conquer those territories? What were their motivations?] ‘Well, I don’t know if there was awareness of Europe and of the continents, but there was awareness that Spain should be a Christian territory [. . .] because it had always been Christian [. . .] The Arabs had forced the Spanish to become Muslims, so that was also a motivation to drive them out again, because they tried to change what their convictions were’. (Marta, 18 years old)
These results highlight the importance and the need to continue fostering historical thinking in the teaching-learning processes of history, thus breaking with the notion of an intuitive historical knowledge understood from common sense. Thinking historically is counterintuitive (Carretero & Lee, 2014), which implies that the student must undergo processes of conceptual change in order to attain a disciplinary understanding of history, much as is the case in the learning of scientific concepts (Carretero et al., 2013).
What narrative would ChatGPT® offer when faced with a request for information on this historical topic so deeply rooted in Spanish national identity? What kind of representation of the past would the student encounter if they were to turn to this AI tool as a source of information? Could these tools facilitate the development of historical thinking in the teaching-learning processes through the opportunities for dialogue it generates?
The following are fragments 1 of a conversation carried out between one of us and ChatGPT® (version GPT-3.5) on the subject of the Reconquista. The approach taken has been exploratory in nature, but we believe that it can reveal fundamental issues in this area. We believe that the educational aims that could be given to ChatGPT should go beyond mere questions and answers. Rather, it should inhabit a context of dialogue and debate, with all the challenges that brings, as well as the enrichment of knowledge-building, not only in the field of history but in many others as well. That is to say, it seems to us that the contribution of ChatGPT in the field of education, inside and outside the school, must be considered in relation not only to how much knowledge it can provide but also to how people can interact with this knowledge in order to construct their own. In this spirit, we transcribe below the dialogue we had, which we have abridged in some cases:
Did the Spanish Reconquista exist?
Yes, the Spanish Reconquista was a historical process that took place in the Iberian Peninsula over several centuries, from 711 to 1492. During this period, the Christian kingdoms in the north of the peninsula gradually fought against the Muslim occupation, recovering territories over time. The culmination of the Reconquista was the capture of Granada in 1492 by the Catholic Monarchs, Fernando de Aragón and Isabel de Castilla, which marked the end of Muslim rule on the peninsula and the establishment of a unified state under the Crown of Spain.
What do you think about the Spanish Reconquista being a myth?
The Spanish Reconquista is a widely accepted and studied term in the history of Spain and the Iberian Peninsula in general. It is not a myth, but a real historical process that took place over several centuries and culminated in the recovery of peninsular territories by the Christian kingdoms [. . .].
In this first interaction we observe how ChatGPT® offers the traditional definition of the Reconquista, denying the possibility that it is an invention and not a true historical process.
Considering the training model of this AI, the answer should not surprise the reader. ChatGPT-3.5® uses as a source of data the combination of vast amounts of texts from the internet, so it is to be expected that its answer fits the popularly accepted knowledge about the Reconquista.
However, in order for this tool to enable the development of historical thinking, the conversation generated should accommodate the use of different evidence or arguments that confirm or deny the information offered. In order to verify this possibility, the following interactions with the chat were oriented to questioning the event, using argumentative elements from Ríos Saloma’s (2005) research.
Did Spain exist before 1492?
Before 1492, in the Iberian Peninsula there were several kingdoms and territories with different cultural and political identities. There was no unified Spain as we know it today [. . .].
Then, how can it be said that the Spanish Reconquista existed in the Middle Ages if Spain did not exist?
You have a valid point. The term ‘Spanish Reconquista’ is a historical convention used to refer to the process of recovering territories on the Iberian Peninsula that were under Muslim rule, there was no unified political entity called Spain. Instead, the peninsula was divided into various Christian and Muslim kingdoms [. . .].
So, although the term ‘Spanish Reconquista’ may seem anachronistic in the sense that there was no Spain as we understand it today in the Middle Ages, it is used retrospectively to describe this historical process in the Iberian Peninsula.[. . .]
However, it is important to keep in mind that the Reconquista was a real historical process that spanned several centuries and had a significant impact on the political and cultural configuration of the Iberian Peninsula. While it has been used as part of the narrative of Spanish identity, it cannot be considered exclusively as a myth, as it is supported by concrete historical events.
A relevant fact for our study arises from this second interaction: ChatGPT® demonstrates the ability to admit the validity of alternative arguments, thus rectifying its discourse. After affirming the non-existence of a unified Spain at the historical moment of the Reconquista, this event goes from being presented as a ‘historical process’ to being presented as a ‘historical convention’.
This response denotes the presence of a substantial element of historical thinking: the ability to understand that historical accounts are not copies of single facts about the past but narratives that are constructed from multiple sources and perspectives. However, as can be seen in the chatbot’s final response, the incorporation of alternative elements did not imply the abandonment of the traditional narrative of the event, reiterating its conception of the Reconquista as a ‘real historical process’ supported by concrete events.
But the idea of Reconquista invents an a priori project that did not exist. So it falsifies history.
You are right to point out that the term ‘Reconquista’ implies a retroactive project that does not necessarily reflect the way people in the Iberian Peninsula in the Middle Ages perceived their own actions. It is a historiographical term that simplifies and frames events specifically from a later perspective.
It is important to recognize that the use of simplifying concepts such as the ‘Reconquista’ in historiography may have limitations and risks. It can lead to a simplistic view of history and distort the understanding of the motives, shifting alliances and complexity of conflicts in that period.
Modern historians often work to better understand and contextualize these historical events, recognizing the limitations of the term ‘Reconquista’ and further exploring the multiple dimensions of the history of the Iberian Peninsula in the Middle Ages. Current historiography tends to focus on a more nuanced and comprehensive view of this period, taking into account the diverse perspectives and narratives that coexisted at the time.
In this last interaction, we observe how ChatGPT® finally abandons the romantic narrative of the event, recognizing the Reconquista as a term that can simplify and distort the historical understanding of what happened in the past. In tune, it demonstrates the ability to examine to what extent the hypotheses developed fit the facts, assuming the complexity of reality and the existence of multiple explanatory arguments for the same event.
The interaction presented suggests that the use of ChatGPT® in educational contexts can be an opportunity to vindicate the importance of dialogue in the teaching-learning processes, opening a space for reflection, questioning and argumentation of first- and second-order concepts in the understanding of historical knowledge. The question, therefore, lies not only in the historical content offered by the tool but in the learning processes it makes possible.
Undoubtedly, there is a risk that the student may encounter false or biased information, but even this possibility can be a potential learning experience in terms of contrasting sources and evidence. To this end, the chatbot should not be the only source of knowledge, in the same way that the textbook in schools should not be the only support for the student to consult. In the field of history teaching and the development of historical thinking, the main responsibility would therefore fall on how the pedagogical strategy incorporating ChatGPT® is planned, as well as on the correct monitoring and accompaniment of the use of this tool.
Impact of AI on knowledge about the past. Challenges
The exploration of how AIs are influencing the teaching-learning processes of history cannot be separated from the challenges posed by their contents in the different spheres of representation of the past. The capacity to generate knowledge on the part of these new technologies raises a necessary concern as to the quality and veracity of historical knowledge.
In the field of popular history, authors such as Shur-Ofry and Pessach (2020) highlight two main challenges at the intersection between AI and collective memory. On the one hand, AI products are not neutral representations of past events but reflect a set of decisions made in their design. On the other hand, these technological products are highly susceptible to manipulations that may impair the authenticity of the content and thus the representation of the past.
These characteristics have also been analysed from a historiographical point of view. Thus, Kansteiner (2022, 2023) offers a critical look at the introduction of ChatGPT® in the work of historians, identifying a number of limitations in the use of this technology.
Outdated information: Since the GPT-3.5 version is based on data available until September 2021, the answers generated may be out of date with respect to recent events or discoveries.
Source opacity: Because the model is trained using a wide range of internet data, it cannot attribute information to specific sources. This raises concerns about the verifiability and reliability of the historical information provided, thus limiting historiographical research.
Principle of sequential plausibility: The probabilistic generation of texts offered by ChatGPT® works under the principle of sequential plausibility, and not of veracity. That is to say, its operation does not seek to offer the most certain answer but the most probable and natural answer to a question from a linguistic point of view.
Censorship of controversial topics: Although the intent of its designers is to prevent the spread of disinformation or offensive content, this may limit the ability of historians to explore and discuss controversial topics in an objective manner.
These characteristics highlight the limitations of this new tool at the disciplinary level: ChatGPT® was neither developed nor trained as an infallible source of academic knowledge, which opens the door to the generation of fictitious or biased content. However, it should be remembered that the limitations mentioned above are not phenomena exclusive to AI. The actions of selecting, manipulating and censoring information are persistent practices in historiography and the transmission of collective memory.
See the case of the Royal Academy of History (Real Academia de la Historia) and the definitions included in its Spanish Bibliographic Dictionary (Constenla, 2011). This institution, dedicated to research and the promotion of productions related to Spanish history, caused a heated reaction by professionals of the discipline as a result of its description of Francisco Franco. In it, his government was identified as an authoritarian regime, but not totalitarian, a nuance that generated a significant academic debate on the rigor of historical knowledge. Likewise, the author’s access to funds from the Francisco Franco Foundation gave rise to numerous accusations about the lack of historiographical diligence in her research. As is well known, Franco, who was the supreme leader of a 39-year dictatorship in Spain, and who won a three-year civil war thanks to the support of Hitler and Mussolini, has been considered a cruel and bloodthirsty dictator by historians of great prestige (Preston, 1993).
The digitalization of the dictionary, as well as changes in the institution’s leadership, brought about a revision of the contents included in the physical version of the work, and with it the redefinition of Franco as a dictator. This event demonstrates that the misrepresentation of the contents of the past is not a phenomenon exclusive to new technologies but rather an inherent aspect of ‘natural intelligence’ and the political use of knowledge about the past. From this perspective, the emergence of artificial intelligences not only poses new challenges but also amplifies and underscores pre-existing risks and challenges in the historical discipline.
Orwell’s (1949) iconic work, 1984, already warned us about the potential of technologies and the control of language in the construction of historical and political reality: ‘Who controls the past controls the future: who controls the present controls the past.’ Intentional manipulation of representations of the past can exert a substantial influence on the way societies understand their present and imagine their future (Carretero, 2018). We are facing a new social challenge that demands the ethical questioning of the mechanisms of control, regulation and use of these new tools. Beyond the educational opportunities that may spring from their use, research from the field of historiography and collective memory is crucial to safeguard a critical approach to the narratives of the past.
Conclusions and final thoughts
In this article we have explored the possibilities and limitations offered by ChatGPT® in the field of history teaching, focusing on its natural language processing capabilities. Through interaction with AI, we have been able to verify that it is possible to generate a space for dialogue in which historical thinking has a place.
The interaction presented around the historical narratives of the Spanish Reconquista suggests that the use of ChatGPT® can be an opportunity to vindicate the importance of dialogue in the teaching-learning processes, opening a space for reflection, inquiry and argumentation of first- and second-order concepts in the understanding of historical knowledge. The question, therefore, lies not only in the content offered by the tool but in the processes it facilitates for the development of historical thinking. For this, an ethical and multidisciplinary perspective is necessary, safeguarding the responsible and beneficial use of these technologies in the construction and transmission of historical knowledge.
Although the comparison between the capabilities of AI and our cognitive abilities is presented as a novel field of study, focusing only on the analysis of the products that these tools generate can lead to reductionism in the discovery of their potential. Therefore, we believe that educational research should not be limited to dichotomizing the understanding of AI and natural intelligence but should also be oriented towards discovering how these technologies can interact constructively, enriching the educational experience of students and allowing them to develop skills in a deeper way.
In a globalized world, overwhelmed by over-information and contaminated by disinformation, a critical approach to knowledge becomes an essential skill for the development of future generations. An interactive approach to questioning the products of AI can not only facilitate the practice of skills specific to the historical discipline but also encourage critical reflection on the information obtained. This possibility should be further explored through empirical research, taking into account the synergies between the different spheres of representation of the past.
In conclusion, the use of ChatGPT® in the teaching of history can make a valuable contribution to the development of historical thinking, as long as its potential and limitations as a pedagogical tool are taken into account. However, new lines of research are necessary to analyse and understand the challenges and risks involved in the generation of historical knowledge by these technologies, thus safeguarding an integral vision of the historical discipline as a phenomenon that runs through the understanding of our past, present and future. Although history is no longer an exclusively human field of knowledge, critical interpretation and the formulation of new questions continue to give us a substantial and irreplaceable responsibility in the preservation, generation and transmission of knowledge.