Mr. Elrod, a Ph.D. candidate in digital humanities at Vrije University Amsterdam, also serves as a lecturer of English at HZ University of Applied Sciences, Zeeland, Netherlands. Although his current research focuses on postcolonial hermeneutics, he is presently interested in LLM generative AI applications in humanities, economics, and artificial general intelligence.  Email: andrew.elrod@hz.nl

 

Abstract

The rapid advancement of artificial intelligence (AI) and large language models (LLMs) has led to the increasing use of AI-generated content in education and research. However, existing citation standards, designed for human-authored content, are insufficient for addressing the unique characteristics and challenges posed by AI-generated content. This article proposes a new citation standard specifically designed for AI-generated content to ensure proper attribution, maintain academic integrity, and promote transparency in its use. The proposed standard includes elements such as AI model name, version, purpose of use, and input prompt. By adopting this citation standard, educators and researchers can effectively integrate AI-generated content into their work while upholding the principles of academic integrity and transparency.

 

Introduction

Shakespeare, in his famous line from Hamlet, likens death to "The undiscovered country from whose bourn No traveler returns." (Act 3, scene 1). Every groundbreaking technological advancement introduces a similar undiscovered country, where old ways die, and new ones emerge. With the rapid advancement of Artificial Intelligence (AI), especially large language models (LLMs) like Google's Bard or OpenAI's GPT-4, we have entered such a territory. Educators and researchers now face the manifold challenges this new reality presents.

The swift advancement of AI has led to an increasing use of AI-generated content, such as text, images, and videos, in educational and research settings. This content offers valuable resources, providing novel insights and perspectives on various subjects. However, the nature of AI-generated content raises questions about academic honesty and transparency. In particular, it is crucial to contemplate how to properly cite and attribute this content in academic work. Traditional citation standards, developed for human-authored content, fail to address the specific characteristics and challenges of AI-generated content.

This article proposes a new citation standard for AI-generated content. This standard aims to ensure proper attribution, maintain academic integrity, and promote transparency in educational and research contexts. The proposed standard considers the unique aspects of AI-generated content, such as the AI model name, version, purpose of use, and input prompt used to generate the content. By providing a clear and consistent method for citing AI-generated content, this standard seeks to facilitate the integration of such content into academic work while upholding academic integrity and transparency principles.

The article begins with a background section discussing the rise of AI-generated content and its implications for education and research. It then explores the need for a new citation standard, highlighting the limitations of existing standards in addressing the unique challenges posed by AI-generated content. The proposed citation standard is presented, along with a sample reference page demonstrating its application in an educational and research context. The benefits and rationale for the proposed citation standard are discussed, emphasizing its importance in promoting academic integrity and transparency.

 

Background

The rise of AI-generated content is attributed to the rapid advancement of artificial intelligence technologies, particularly in the field of natural language processing (NLP). NLP models, such as OpenAI's GPT, have demonstrated remarkable capabilities in generating human-like text, making it increasingly difficult to distinguish between human-authored and AI-generated content. This has led to a growing interest in using AI-generated content in various domains, including education and research.

In educational settings, AI-generated content can be used to supplement teaching materials, provide personalized learning experiences, and generate new insights and perspectives on various topics. For example, AI-generated content can be used to create customized textbooks, generate practice questions, or provide explanations and examples tailored to individual students' needs. In research settings, AI-generated content can be used to generate hypotheses, analyze data, and synthesize information from multiple sources, potentially accelerating the pace of scientific discovery.

However, the increasing use of AI-generated content in education and research also raises concerns about accuracy, academic integrity, and transparency. Since AI-generated content is not authored by a human, traditional citation standards do not adequately address the unique characteristics and challenges posed by this type of content. For example, existing citation standards typically require the author's name, publication date, and source, but these elements are neither applicable nor sufficient for AI-generated content.

Furthermore, the use of AI-generated content in education and research can also raise ethical concerns, such as the potential for AI-generated content to perpetuate biases, misinformation, or disinformation. As a result, it is crucial to develop a citation standard that not only ensures proper attribution of AI-generated content but also promotes transparency and accountability in its use, as well as providing a means of replication and critique.

 

The need for a New Citation Standard

The need for a new citation standard for AI-generated content arises from the unique characteristics and challenges posed by this type of content. Traditional citation standards, which were developed for human-authored content, do not address these challenges, leading to the aforementioned concerns.

One of the main limitations of existing citation standards is their focus on human authorship. Traditional citation standards typically require information such as the author's name, publication date, and source, which are not sufficient for AI-generated content. For example, AI-generated content does not have a human author, as the content is generated on demand. In this way, the traditional role of author has become that of prompter and/or collaborator. Moreover, the source of AI-generated content is the AI model itself, which is not identifiable or citable using existing citation standards.

Another challenge posed by AI-generated content is the potential for biases, misinformation, or disinformation. Since AI models are trained on large datasets, they may inadvertently learn and reproduce biases present in the training data. Furthermore, AI-generated content may not always be accurate or reliable, as the AI model may generate content based on patterns and associations in the training data rather than factual information (i.e., hallucinations). As a result, it is crucial to develop a citation standard that not only ensures proper attribution of AI-generated content but also provides transparency in prompting technique.

Additionally, the dynamic nature of AI-generated content presents further challenges for citation. AI-generated content can be generated on-demand and may vary depending on the input prompt and other factors, making it difficult to consistently cite the same content. A new citation standard for AI-generated content should take into account these unique aspects and provide a clear and consistent method for citing AI-generated content in educational and research contexts.

 

Proposed Citation Standard

The proposed citation standard for AI-generated content aims to provide a clear and consistent method for citing such content in educational and research settings. The standard will include the following elements:

AI Model Name (Version). (Year). Purpose of use. Retrieved from [URL or DOI], Prompt: [Input prompt].

Each element plays a crucial role in ensuring proper attribution and recognition of AI-generated content, promoting transparency and accountability in the use of AI models.

The model name is essential for identifying the specific AI system that generated the content. Different AI models may produce varying results, and citing the model name allows readers to understand the origin of the content and assess its credibility.

The version of the model is another important element, as AI models are frequently updated and improved. Including the version number in the citation ensures that readers can track the evolution of the AI model and understand the context in which the content was generated. The designation of a date or year is also necessary as some AI models (e.g., GPT-4) will have varying abilities across time within the same version (e.g., consider the ChatGPT version 4 May 3, 2023 instance in contrast to the March 23 instance).

The purpose of use is a critical component of the citation standard, as it provides context for the AI-generated content. By stating the purpose of use, readers can better understand the intended application of the content and evaluate its relevance to their own research or educational needs. Additionally, citing the purpose for using the technology promotes transparency and contributes to academic honesty.

Finally, the prompt used in generating the AI content is an essential element of the citation standard. The prompt serves as the input for the AI model and can significantly influence the output. Including the prompt in the citation allows readers to understand the basis for the AI-generated content and assess its validity and relevance. As well, this is essential for the replication of research results and/or critique of the material generated.

By incorporating these elements into the proposed citation standard, we can ensure proper attribution of AI-generated content, promote transparency and accountability in AI use, and facilitate communication and collaboration in research and education.

 

Sample Use of AI Reference Page

OpenAI GPT-3 (3.5). (2022). Educational material generation. Retrieved from https://www.openai.com/gpt-3, Prompt: 'Explain the process of photosynthesis in simple terms.'

OpenAI ChatGPT (4). (May 3, 2023). Research summary. Retrieved from https://chat.openai.com, Prompt: "Summarize the main findings of the article titled 'The Impact of Climate Change on Biodiversity'."

This sample reference page demonstrates how the proposed citation standard can be applied to AI-generated content in an educational and research context. The citations include all necessary information as previously discussed.

 

Benefits and rationale

The proposed citation standard for AI-generated content offers several benefits and addresses the unique challenges posed by this type of content. The main benefits and rationale for the proposed citation standard include proper attribution, transparency, compatibility with existing citation styles, and adaptability to future developments in AI-generated content.

Proper attribution is a key benefit of the proposed citation standard. By providing a clear and consistent method for citing AI-generated content, the citation standard ensures that the AI model responsible for generating the content is properly credited. This is particularly important in educational and research settings, where accurate attribution is essential for maintaining academic integrity and fostering a culture of intellectual honesty.

Transparency is another important benefit of the proposed citation standard. By requiring information about the AI model, version, purpose of use, and the prompt used, the citation standard promotes transparency in the use of AI-generated content. This can help address concerns about biases, misinformation, or disinformation, as users can better understand the origins and limitations of the AI-generated content they are using or citing.

Compatibility with existing citation styles is a key rationale for the proposed citation standard. The citation standard is designed to be easily integrated into existing citation styles, such as APA, MLA, and Chicago, ensuring that users can continue to use their preferred citation style while incorporating AI-generated content. This compatibility also facilitates the adoption of the citation standard in educational and research settings, as users do not need to learn an entirely new citation style.

Finally, adaptability to future developments in AI-generated content is an important consideration in the design of the proposed citation standard. As AI technologies continue to evolve and improve, the citation standard should be flexible enough to accommodate new AI models and use cases. By providing a clear and consistent method for citing AI-generated content, the citation standard can help ensure that the academic community is well-positioned to address the unique challenges and opportunities presented by AI-generated content in the future.

 

Conclusion

Some bridges can only be crossed once.  The technological disruption brought about by AI-generated content is equivalent to the advent of the Internet and introduces an undiscovered country for education and research.  As this content becomes increasingly prevalent in education and research, it is essential to develop a new citation standard specifically designed for this new reality. By adopting this citation standard, educators and researchers can effectively integrate AI-generated content into their work while upholding the principles of academic integrity and transparency. The academic community must embrace the opportunities presented by AI-generated content while being mindful of the challenges and ethical considerations that accompany its use. The development and adoption of this citation standard is a crucial step towards ensuring that the integration of AI-generated content in education and research is conducted responsibly and transparently.

As AI-generated content continues to advance, it is essential for the academic community to stay informed and engaged in discussions surrounding the ethical use of AI technologies. This involves not only the adoption of citation standards but also the examination of potential biases, misinformation, or disinformation that AI-generated content may introduce. By maintaining an ongoing dialogue and being proactive in addressing these concerns, educators and researchers can ensure that the integration of AI-generated content in education and research remains beneficial and productive.

Ultimately, the new citation standard proposed in this article is just one component of the broader effort to adapt to the rapidly changing landscape of AI-generated content. As AI technologies continue to evolve, so too must our approaches to understanding, using, and citing these tools in educational and research settings. By staying informed, engaging in discussions, and adopting best practices like the proposed citation standard, the academic community can ensure that the healthy integration of AI-generated content into education and research is carried out responsibly, ethically, and transparently.

 

Please check the Pilgrims f2f courses at Pilgrims website.

Please check the Pilgrims online courses at Pilgrims website.

 

 

References

“Introducing Google Bard.” n.d. bard.google.com. https://bard.google.com/.

“ChatGPT — Release Notes.” n.d. help.openai.com. https://help.openai.com/en/articles/6825453-chatgpt-release-notes.

OpenAI. 2023. “GPT-4.” openai.com. 2023. https://openai.com/product/gpt-4.

Shakespeare, William. 2015. “Hamlet: Act 3, Scene 1.” MyShakespeare. December 27, 2015. https://myshakespeare.com/hamlet/act-3-scene-1.