In late 2024, the EERA blog editors invited the Editorial Team and members of the EERA family to join an online discussion on the use of AI on the EERA blog. Our goal was to provide advice on the appropriate and ethical use of AI in academic writing, and to create AI guidelines for blog contributors.
We want to work with our contributors to create excellent blog posts that our readers will enjoy. We emphasise that these are guidelines, not rules or prohibitions.
Using AI in academic blogging
AI and academic writing – the importance of the process
Writing isn’t just about hitting your word count. During the process of writing, you engage deeply with ideas, question assumptions, and refine your arguments. You aren’t just writing to create excellent academic content, but to develop academic skills and knowledge that will help you in your future career.
You can use AI to create drafts, summaries, and even generate ideas that might take you down a different path, but AI doesn’t think. It only reproduces. It does not have the capabilities to develop original ideas or conclusions – that’s your job as a researcher.
Finding your voice
One of the exciting things about writing for an academic blog is that it can help you develop your unique academic voice, where your thoughts, your perceptions and your insights are expressed in your own way.
While the academic writing style is more formal and impersonal, it doesn’t have to be boring or bland, especially when you are writing a blog post. But that’s what happens if you ‘outsource’ your writing to AI, because, of course, AI doesn’t have a voice, and it doesn’t have your creativity.
Fact-checking – AI and accuracy
ChatGPT and other AI services are based on a data set that may not include the latest research. In addition, AI has been known to generate information that sounds plausible, but which, on closer inspection, is revealed to be incorrect – a phenomenon known as ‘hallucination’. At times, when asked for a source, AI tools have hallucinated researchers, papers, and PubMed ID numbers.
It is absolutely critical that you fact-check everything that AI suggests, including any data, references, or information. Be cautious with AI summaries of academic papers – sometimes these are accurate, but they may also pick up on an aspect that wasn’t the main focus of the paper.
Copyright, ethical concerns, and academic integrity
AI models are trained on large datasets that include copyrighted material. It is almost impossible to determine exactly where the AI tool has sourced the information, which would leave the author open to a charge of copyright violation.
While the legal implications of using AI are not yet clear, concerns around ethics and academic integrity are being discussed among scholars and educational institutions. Is your work original and accurate, or are you passing the work of others off as your own?
AI in academia and diversity
AI can provide assistance for multilingual students and researchers, or those with disabilities. However, users should keep in mind that AI algorithms may exacerbate racial disparities in education, and students from low-income backgrounds may not have access to generative AI tools, and the advantages they offer.
Sustainability concerns
In keeping with EERA’s policy on sustainability, we must also consider the effect of increasing AI usage on energy consumption and natural resources.
Advantages of AI in academic blogging
Despite our advice on being cautious when using AI in academic blogging, we also want to recognise that there are advantages.
We’ve all sat in front of an empty page and struggled to start writing. AI can provide a starting point or a rough outline that helps in the initial drafting process. It can suggest related topics or concepts or link them in a way you may not have considered.
Generative AI can increase equity in academia, for example, with language tools that help non-native English speakers, and speech-to-text tools and text-to-speech tools for those with physical or visual impairments. We recognise and welcome that non-native English speakers who wish to publish an article on the EERA Blog can use AI translation tools or proofreading tools to improve their text – giving them the confidence to submit their work.
Our AI guidance provides transparent and practical advice to EERA contributors on the use of generative AI tools.
EERA AI Guidance – Acceptable use of AI for EERA blog contributors
Brainstorming ideas
AI can be incredibly useful for generating topic suggestions, or exploring links between concepts. It accesses sources that you might not think of, and can encourage you to go beyond your familiar perspective. Out of your comfortzone, so to speak.
Just watch out that it doesn’t lead you way off track. You are the one to decide whether the idea that AI generates fits with the theme and argument of your blog post, and which is the most pertinent.
Researching sources, finding connections, analysing data
Some AI tools can help students find initial sources, gather and format references, or find connections or patterns.
You can also use AI to summarise papers to see if they are relevant to your work – though you should, of course, take the time to read the paper if you want to refer to it. Perhaps you will discover an aspect that AI didn’t highlight in the summary, that takes you in a different direction.
Narrowing the scope of a topic
If you are writing about your research, it can be difficult to narrow the focus – especially if the blog post should be around 700 – 1000 words. You might struggle to summarize your years of research and knowledge into an interesting yet concise blog post that doesn’t overwhelm the reader.
Feed your ideas into an AI tool and ask it to suggest one aspect of your work on which to concentrate. You can always submit another blog post in the future, that could look at other aspects.
Drafting an outline
When you start writing, it can be helpful to have a rough outline of the topics you want to cover, and what you want to say. AI tools can you organise your thoughts before you start the actual writing process.
With this, we are talking about what you will write in each section of your blog (e.g., the introduction, background information, arguments/takeaways, summary, key messages), not a rough first draft of your post (see below).
Suggesting a headline or subheadline
Sometimes creating a good headline or subheadline can take a long time. We find that academics are used to long, sometimes complicated headlines. This may be fine on an academic paper, but on a blog, the headline needs to grab the reader’s attention and encourage them to engage. On a practical note, they also don’t fit into our blogging software.
We encourage the use of subheadings, which break up the text and make it easier for readers. AI is actually pretty good at summarising a blog post and suggesting headlines.
Creating an illustration
This is an area where the results still vary widely, but AI can be used to create an illustration to visualise your data, or for the featured image on your blog post. Particularly with illustrations and images, you should be careful not to use copyrighted source images. This is an area that is quite fraught at present, and it is sometimes difficult to identify the original source.
Full disclosure here – we have tried using AI to create our featured images, but the results were so unsatisfactory that we never used them.
Editing or proofreading text
This is incredibly useful for authors whose native language isn’t English, or for those who have a learning disability such as dyslexia. Expressing complex academic concepts clearly in your second (or even third!) language can be challenging.
AI tools can help check your grammar, improve sentence structure, and suggest alternative wording. Do be careful here though, that you don’t just blindly accept all suggestions. Even excellent tools such as Grammarly can be wrong, or overly prescriptive, intruding on your style.
Reducing the editorial workload of co-authors and ensuring consistency
When writing a collaborative academic blog post, you might not want to ‘burden’ the one native speaker on the project with editing the contributions of the others (even if they say they are ok with it!)
Using an AI proofreading tool can spread this work between all participants. It can also helps ensure consistency in language and tone.
The following uses of AI are not acceptable on the EERA Blog
Writing a complete blog post
“Why bother reading something that no one was bothered enough to write”
There’s a lot of truth in this quote. If your research is good enough to write about, it’s good enough for YOU to write about it. And no one can write better about your research than you. Not even AI.
You have the opportunity to put your research in front of the EERA community, and reach educational researchers throughout the world.
Why waste this opportunity with a generic AI blog post? Let your voice shine through!
Creating a ‘first draft’
We came across this in our research. It is often presented as an acceptable use of AI. We disagree. There’s a danger that when you ask AI to create a first draft, that you will think, ‘That’s not bad’, and use more of it than you should. You might also become blind to alternatives, because you’ve been steered in a specific direction by AI.
If you rely too much on AI for your first draft, you’ve skipped the first part of writing an academic blog – as we mentioned, the creative process where you develop original ideas. You’ve missed the opportunity to analyse sources, investigate ideas, and express these in your own voice. That’s what makes a blog post interesting and thought-provoking to read.
Restating a sentence or a paragraph
On a similar note, if you feed a list of your thoughts into the AI tool and ask it to restate your text, or restructure your paragraph, this is not your own work. This isn’t the same as using an AI tool for editing text that you have written – in that case, you are still using your own words.
In addition, there is a risk that the AI tool could misinterpret what you really wanted to say, and you might not spot the mistake. Assessing a text at this level may be especially difficult for a non-native speaker.
Finally, having a tool ‘restate’ your unstructured notes will result in a text that feels generic and inauthentic – and again, it will not reflect your original voice.
Did you know – our blog editors Lynn and Millie try to have a light hand when editing posts by non-native speakers. Our goal isn’t to produce blog posts written in ‘perfect’ English, but to let the voice of the author shine through.
Where a German speaker is more likely to write in the passive voice, and use long, complicated sentences, an Italian or French speaker might use more expressive language. The EERA family is international – we want our blog to reflect and celebrate that diversity.
Millie speaks English, Welsh, Greek, Italian, and French. Lynn speaks English and German. We are happy to help if you have a translation question.
The EERA editing process
We will clearly disclose any AI tools we use for editing and providing feedback to authors, for posting on the blog, and for sharing your posts on social media. At present, we use a proofreading tool (Grammarly) for a final check on spelling and grammar.
Disclosure
Transparency is key. We ask that all contributors provide clear disclosure of all tools that were used in the writing process, and the extent to which these were used. This includes all tools used to research the topic, generate content (text and images), and proofread text. Contributors bear full responsibility for the article’s accuracy and validity, and that the content (including images created with AI) is not copyrighted.
If we suspect that an AI tool has been used without disclosure, we will reach out to you for clarification. We may ask you to rewrite or rework your submission. We do not use AI detection tools, as they have proven to be unreliable, and even biased against non-native English speakers.
Example of disclosure:
ChatGPT was used to research a question on the use of AI in education, and to suggest a blog structure. Grammarly and Deepl were used for grammar, spelling, and style check of the blog posts. Most of the corrections and suggestions were accepted.
The use of AI translation tools
Non-native English speakers may submit work that has been translated using an AI tool, which will be edited by our team, in cooperation with you. In this case, please also submit the original language text, as this can help us identify words or concepts that may have been ‘lost in translation’.
How universities around the world see AI in academic writing:
Ethical Use of AI in Writing Assignments, University of Kansas, USA https://cte.ku.edu/ethical-use-ai-writing-assignments
Guidelines for Using AI Tools in Writing and Research, Walden University, USA https://academics.waldenu.edu/artificial-intelligence
Guidelines for responsible AI use in academic writing, Stellenbosch University, South Africa https://www.aidscentre.sun.ac.za/guidelines-for-responsible-ai-use-in-academic-writing/
AI and academic writing, Örebro University, Sweden https://www.oru.se/university-library/support-for-students/guide-to-academic-writing/ai-and-academic-writing/
Using Generative AI, Universitiy of Cambridge: https://blendedlearning.cam.ac.uk/guidance-support/ai-and-education/using-generative-ai
Artificial Intelligence and Your Learning, Newcastle University, UK: https://www.ncl.ac.uk/academic-skills-kit/good-academic-practice/artificial-intelligence/
Tracy A. Mendolia has compiled a Padlet of University Policies on Generative AI and this paper The global landscape of academic guidelines for generative AI and Large Language Models also provides an overview of policies and guidelines.
Additional information that we found interesting:
In this article on AI and creativity, the authors showed a generally positive response to the use of using AI, but caution that some students found they sometimes got stuck on the AI idea, and found it difficult to come up with their own ideas. The researcher also notes that as AI recycles existing content, the students found the AI suggestions repetitive overall – even if their individual creative output improved.
Practical guidance for leaders when developing guidance for their school systems, from the World Economic Forum sets out seven principles for AI in education – Purpose, Compliance, Knowledge, Balance, Integrity, Agency, and Evaluation.
In ‘How Writing Leads to Thinking’, historian Lynn Hunt notes, ‘Writing means many different things to me but one thing it is not: writing is not the transcription of thoughts already consciously present in my mind. Writing is a magical and mysterious process that makes it possible to think differently.
This article warns against using AI Detectors, and provides advice on what to do instead, while this paper shows that ‘false positives disproportionately affect non-native English speakers and scholars with distinctive writing styles’.
In this paper on the impact of cognitive offloading, researchers found ‘a significant negative correlation between frequent AI tool usage and critical thinking abilities, mediated by increased cognitive offloading’.
In The Negative Impact of AI on Academic Integrity in Tertiary Education, the authors warn that, ‘Students who rely on AI without engaging in genuine learning may achieve academic success yet remain ill-prepared for the demands of their respective industries’.
This article in Nature, AI firms must play fair when they use academic data in training, includes a call for research into new kinds of licenses, or changes to copyright laws ‘Generative AI tools are using a data ecosystem built by open-source movements, yet often ignore the accompanying expectations of reciprocity and reasonable use, says Sylvie Delacroix, a digital-law scholar at King’s College London.’
The EU article on Artificial intelligence and copyright: use of generative AI tools to develop new content, provides an overview of the current and future legistlation, and has a checklist of aspects to consider when evaluating the use of AI software to generate new content.
In this blog post, Copyright, Education, and Generative AI: Getting with the programme?, the authors provide an overview of GenAI policies in UK universities, and discuss questions around AI use by students and copyright concerns.
Are your students aware of your AI guidelines?, ask the authors of this blog, stating that ‘only 5% of students are aware of their university’s AI guidelines and feel they are fully comprehensive’.