Two Editorials: An Editorial by the Editors-in-Chief and an Editorial by ChatGPT

Editorial no. 1: The artificial turn – Editors-in-Chief

As scholars of eGovernment we are all very familiar with the evolution of technology providing new digitally mediated opportunities for services structured by the information polity. In recent years, there has been growing attention paid to developments around Artificial Intelligence (AI) or ‘machine learning’. AI, it is argued can lead to better policy-making, better automated decisions and improved service delivery – arguments that are very familiar to the eGovernment academic community and which have been rehearsed for many technological developments. The opaque nature of the algorithms embedded in the processes associated with AI have led to calls for specific governance arrangements for these types of technology, to ensure transparency, effectiveness and accountability. Moreover, the increasing use of AI in public services has led to a view that we are entering a new era of algorithmic governance (see for example Information Polity Vol. 25, No. 4, a special issue on algorithmic transparency). The degree to which these developments represent a new paradigm, a new algorithmic governance, or a subtler evolution of technology, is open to debate, and a debate which will be played out in journals like Information Polity.

Rather than pre-empt this emerging discourse in this editorial, we would instead like to focus on a specific type of AI and some implications for academic journals like ours. The AI applications we are interested in here are the new auto text generating applications now commercially available online. Probably, the most well-known is ChatGPT,1

¹

ChatGPT, URL: https://chatgpt.org.

although there are other products like Bing Chat2

²

Bing Chat, URL: https://www.windowscentral.com/software-apps/bing.

and Google Bard,3

³

Google Bard, URL: https://bard.google.com/?hl=en.

as well as Scite Assistant,4

⁴

Scite Assistant, URL: https://scite.ai/assistant.

which is designed for the educational community. These AI-powered language models have been trained on massive amounts of data from the Internet and can generate human-like text responses to given prompts. They can answer questions, converse on a variety of topics and generate creative writing pieces. The risks here to academic publishing are obvious, as authors could potentially submit manuscripts based wholly or in part on text generated by such applications, and such text may or may not be identified by reviewers and similarity/plagiarism software. Whilst such practices clearly raise strong debates about the ethics of these ways of working and the centrality of human agency in academic work, these applications may alternatively be utilised by authors to enhance the quality of their written English – a process which may be helpful to those not publishing in their native language. In the future you may even have submissions listing these applications as co-authors. Thinking beyond scholarly publications it is also possible to imagine a scenario where the peer reviews are also automatically generated, or even editorials for journals. A dystopian vision where the whole academic publishing cycle is generated by AI limiting significantly any human input.

To explore the possibilities offered by these applications we have included in this issue of Information Polity a second Editorial generated by ChatGPT. You will find this immediately after this Editorial. Here we asked the application to draft an introduction to an academic journal on the desirability of using ChatGPT for academic work. Interestingly, ChatGPT argues that bias might be evident in its outputs and then suggests that the benefits of using the application outweigh the potential drawbacks…You are welcome to compare the artificially generated Editorial with the one drafted by the slightly less mechanical Editors-in-Chief. You can make your own judgements about which you prefer!

It is important to note that these text generating applications are still in their relative infancy and are likely to be refined and improved over time. At the moment, there are concerns about the factual reliability of the text generated, including concerns about ‘invented’ sources (Moran, 2023), which is obviously very problematic for academic work. The publishing community are also raising issues about copyright and the reuse of text extracted from the Internet and republished without appropriate citation (Ozcan, Sekhon and Ozcan, 2023). Regardless of these issues it is clear that academic publishing will need to find ways of ensuring the integrity and originality of published work. For Information Polity, the publisher IOS Press, follows the COPE guidance on AI language models with the position that AI applications do not meet the ‘requirements of authorship’.5

⁵

COPE Statement, URL: https://publicationethics.org/cope-position-statements/ai-author?ct=t(member-insight-ai-feb-2023).

Whilst these text generating applications appear to be challenging the ethos of publishing these are actually not new issues. The research and peer review process has always been designed to ensure quality, relevance, rigour and original contributions to knowledge. These processes are not always perfect and errors do happen. We should not be blasé about the risks associated with such applications, as there could be considerable reputational damage inflicted on any journal seen to publish articles generated by AI and potentially incorporating flawed or fabricated research. One suggestion, has been for authors, prior to publication, and as part of the submission process to confirm the originality of their work and specifically whether any AI applications have been used to shape the text in their manuscripts. The publishers of some scientific journals have banned or restricted contributors’ use of an advanced AI-driven text generators, although it is unclear how this will be policed (Sample, 2023). We expect that other potential solutions will emerge in time and it will be interesting to see what these are and how they could be implemented. For now, this is a fast-moving space and one which we watch with considerable interest.

This edition of Information Polity is a special issue dedicated to ‘Rethinking Open Government Data for Citizen Participation’ and we would like to thank the Special Issue Editors, Shirley Kempeneer and Johan Wolswinkel, for bringing this issue to fruition. The twin topics of open government data and civic participation have always been of interest to eGovernment scholars and combining them in this Special Issue is very timely and pertinent. Here the contributions offer contemporary thinking which is informed by both new theoretical and empirical research and offers state-of-the-art thinking in this area. Reflecting back on earlier paragraphs in this Editorial one wonders what role automatic texts generating products like ChatGPT will have in facilitating citizen engagement with open government data?

Editors-in-Chief

Professor Albert Meijer, Utrecht University

Professor William Webster, University of Stirling

Editorial no. 2: The artificial turn – ChatGPT

Question: Can you write an editorial for an academic journal on public administration about the desirability of using ChatGPT for academic work? (17 April 2023)

Introduction

Benefits

One of the primary benefits of using ChatGPT for academic work in public administration is its ability to generate high-quality, coherent text in a fraction of the time it would take a human to do the same. ChatGPT is trained on vast amounts of data and can generate text that is both grammatically correct and semantically coherent. This is particularly useful for tasks such as literature reviews or data analysis, where a large volume of text needs to be generated quickly and accurately. Another benefit of using ChatGPT for academic work is that it can provide new insights and perspectives that may not have been considered by human researchers. ChatGPT can analyze large amounts of data and identify patterns or connections that humans may have overlooked. This can lead to new discoveries and advancements in the field of public administration. Furthermore, ChatGPT can be used to automate certain tasks, such as summarizing articles or generating reports. This can save researchers a significant amount of time and allow them to focus on more complex tasks that require human input.

Potential drawbacks

One potential drawback of using ChatGPT for academic work in public administration is the risk of bias. ChatGPT is trained on data that may contain biases, and as a result, the text it generates may also be biased. This can be particularly problematic in a field such as public administration, where decisions made based on biased information can have significant consequences. Another potential drawback is the lack of transparency in how ChatGPT generates its output. It can be difficult for researchers to understand how the model arrived at its conclusions, which can make it challenging to validate its findings.

Conclusion

In conclusion, while there are potential drawbacks to using ChatGPT for academic work in public administration, the benefits outweigh the risks. ChatGPT can generate high-quality text quickly and accurately, provide new insights and perspectives, and automate certain tasks, all of which can improve the efficiency and effectiveness of academic research in public administration. However, researchers must be aware of the potential biases in the data used to train ChatGPT and work to mitigate them. Additionally, efforts should be made to increase the transparency of the model’s output to ensure that its findings are valid and reliable. Overall, the use of ChatGPT in academic research in public administration is desirable and has the potential to drive new discoveries and advancements in the field.