.jpg)
by David Emanuel Elcock | Laidlaw Scholar (2025) at Trinity College Dublin, Ireland for Localise Youth Volunteering at Claremont Avenue, Glasnevin, Dublin, D11 YNR2
Return to the Home page: Climate Justice Volunteering Toolkit | Homepage
When partaking in any form of activism or volunteering it benefits having a solid understanding of the issues and problems that are attempted at being addressed. Knowing the common pitfalls and strategising ahead of them can drastically improve the success rate of projects. Furthermore, it leaves participants with a greater confidence going into the project and a greater sense of having gotten something out of it. The path to acquire this information is by conducting research and on this page a number of topics related to conducting research are discussed:
Artificial Intelligence (AI) like any tool is only effective when yielded properly. Whilst it can be good in summarising research and provide a coarse overview, it is advised to avoid relying on AI as a primary resource. This is because of the existing pitfalls with AI:
The effect of these problems can be minimised by improving prompt writing and managing reasonable expectations. For example, if you request an AI model to write you a 2000 word essay on a given prompt the AI model will write a poor quality essay. A better output can be generated by providing the AI with details on the course such as lecture notes. Furthermore, by asking the AI to write the introduction in one prompt, the body in a second and the conclusion in a third, each part is written at a greater quality as more of its computation per prompt is allocated to a single task. More context and asking for less per prompt, spreading tasks over multiple prompts is more reliable yet remains imperfect.
Below are some questions to consider when identifying resources:
When looking for research materials it can be useful to utilise a search engine instead of visiting specific journal for relevant materials. There are many search engines beyond a Google or social media search:
When unfamiliar with a research landscape it can be useful to make use of these search engines. Typically when a scholar becomes more familiar with the research landscape (experience and exposure) they will focus on reputable academic sources, often specific journals they know to be highly influential and up to date.
When researchers write up the findings of their research they try to have it published as a paper. A draft manuscript is sent to a journal for peer review, usually the highest one they think will accept their work. There are two types of journals:
It should be noted that this model applies to publicly funded research hence a good amount is left behind a paywall from the public. In order to gain access, research institutions thus often have large contracts with these journals to give access to researchers and students. But, in physics a work-around has been introduced: By convention all papers are first uploaded as a pre-print on Arxiv. Hosted by Cornell University this platform accepts papers that before they are peer reviewed. Upon peer-review the authors typically update the manuscript with a link to the publication, but the paper remains accessible for readers via Arxiv. It might be something other academic fields should consider.
The reason the publication process is discussed here is to illustrate a problem that caught society’s attention during Covid-19, not all the publications are peer reviewed. Without peer review the paper is seen as less reliable as it hasn’t been screened for errors. Hence whenever a paper is being sourced, make sure it isn’t a pre-print and that it has in fact been published in a journal.
In 2021 the government of Greenland and Denmark were shocked to learn that scientists had measured high levels of mercury in the territory of Greenland. This is because Mercury is a toxic metal which leads to multiple serious diseases in humans. However, in 2024 a new study concluded that the 2021 findings were invalid and possibly due to a measurement contaminant.$^{[1]}$ When researching topics it helps to consider that citations can be invalidated over time and so it requires the researcher to check for updates of papers, topics and fields.
A lot of emphasis has been placed on papers in part due to the peer review, novelty and abundance. But other sources to consider are:
The rise of AI has exacerbated a problem in academia that has existed and probably will persist long into the future: academic slop. With this I refer to the phenomenon of researchers purposefully producing junk material. Why you might ask? Because in academia there exists several metrics e.g. the h-index that quantify how good of a researcher you are by looking at the number of publications a scholar has and how often that paper has been cited by others. Having a higher h-index is considered as proof that the researcher has high performance which increases their access to research grants, career progression and secondary opportunities like invitations to be invited as a speaker at conferences. There is thus an interest to game the h-index and paper mills pave the way for that to happen.
For as little as 50 EUR, you can approach dodge companies to have you name added to a paper you never worked on that is published in a reputable journal. What is scary is that we do not know the extend of this problem. It is therefore paramount to become comfortable in critical thinking and asking yourself if the results actually make sense and if they can be verified in other ways e.g. reproducibility.
It should come to no surprise that two researchers don’t exactly agree on a measurement. The exact mass of a chicken will vary depending on the location, age, nutritional intake, etc. But if you see that a large difference is reported, it might indicate that at least one of the values is less trustworthy. However, you would only discover this problem upon acquiring at least two independent references that agree on a reasonable range.
What is reasonable will depend on the subject area. Perhaps getting a precise measurement is challenging or different methods having a different precision. What matters is agreement. If the estimates give different scales in the same unit, you cannot reliably reference a fixed number. It is possible however to include both references and point out the disagreement in the literature.
There is a multitude of referencing styles used in academia. In some cases a field will prefer a common style, sometimes a journal will insist on their format and other times again the author is free to choose. A universal requirement however is that the referencing style chosen remains fixed throughout the research you conduct.
An extensive overview on some referencing styles and which subject prefers what format can be found here: *LibGuides at UCD Library.* The School citing Styles tab allows users to find out what referencing style is appropriate for their subject.
If you wish to streamline the process, Referencing Management Softwares such as *Mendeley, Zotero & Endnote* can be considered. In specific text editor software like Latex these can be scripted to update dynamically and be given names so that you don’t have to remember which number goes with what reference.
[1] C. J. Jørgensen, J. Søndergaard, M. M. Larsen, K. K. Kjeldsen, D. Rosa, S. E. Sapper, L-E. Heimbürger-Boavida, S. G. Kohler, F. Wang, Z. Gao, D. Armstrong, C. N. Albers, Sci. Adv., 26, 10, 4, 2024.
In the era of Artificial Intelligence (AI), when conducting research it is increasingly important to extend beyond the final product and actually document how research is conducted. Good research habits are most reliably obtained by practice and skills via analysing approaches. With access to AI, teaching good research practices has generated headaches in academia as staff are unsure how to proceed with offering fair assessments for students. Determining how effective the student’s conducted research is becomes challenging in for example assessing how AI was used in their work. E.g. did they only use AI for literature review summaries or also for drafting their work?
As an illustration to the concerns in academia, a famous scandal broke out in 2024 when three Chinese scholars were caught having used AI in generating academic slop and tried to have it published. To everyone’s horror, the succeeded to publish their work in *Frontiers in Cell and Developmental Biology,* a high impact and reputable journal. The paper went viral after peers identified serious anatomical inconsistencies and terminology. This shows the concerns by academics, trying to ensure this isn’t normalised by students.
A modern view of the challenge of AI is to respect the problem as falling under the ‘wicked’ category.$^{[1]}$ Features of ‘wicked’ problems include: (not limited to)
By recognising the problem as ‘wicked’, instead of seeking a perfect all-fit solution, the mindset shifts to strategies that can cope with compromise, diverge in approaches and iterate over time. Since conducting research is part of assessment, it stands to reason the same approach will be required in academia by researchers. For example, the preprint server ArXiv for physical sciences announced in 2026 that penalties will be introduced on researchers uploading papers that contain AI hallucinations. These penalties include a year full ban from uploading their papers and a life-long requirement of papers having been submitted to reputable peer-reviewed venues, effectively a ban on pre-prints.$^{[2]}$ For researchers whose career are under a publish or perish stress this demonstrates the need for strategies that enable students to learn to use AI, be assessed in these methods effectively and for researchers to adapt to the use of AI responsibly.
One suggestion for adapting is to make us of the Black Box Assessment method which emphasises the importance of the process of research and suggests documenting this like a black box flight recorder.$^{[3]}$ Similar to investigations into aviation tragedies the black box would become a tool to assess learning thus provides opportunities for feedback and optimisation. Students could be introduced into the habit of documenting elements of research which offer ‘windows’ into the input to output trajectory. Examples of these ‘windows’ include: (not limited to)
The advantages of the Black Box Assessment method is that it is structural, it addresses how assessments are done instead of being discursive i.e. about imposing limitations.$^{[4]}$ As such it could offer a constructive approach to prepare youth volunteers from disadvantaged backgrounds into their volunteering mission as well as into their academic interest by including Black Box Assessment into their volunteering routines. Obviously any serious and impactful climate justice volunteering will require members to be familiar with a critical knowledge of relevant search topics hence it is a vital skill to be comfortable to reliably conduct this research. It is also important for people to become comfortable with AI both in using it and recognising when it is used. For example, can you answer Q5 below?
I made this slide as part of an event at the Physics Society at Trinity College Dublin and the failure rate in identifying which picture was real and which was AI was surprisingly high!
[1] T. Corbin, M. Bearman, D. Boud & P. Dawson, Assess. Eval. High. Educ., 2025, p. 1-17. [2] D. S. Chawla, Nature News, 2026. [3] N. M. Winstone, K. Gravett & S. Elkington, Assess. Eval. High. Educ., 2026, p. 1-18. [4] T. Corbin, P. Dawson & D. Liu, Assess. Eval. High. Educ., 2025, 50, 7, p. 1087–1097.
Taking inspiration from the guidelines to conducting research, advise and experience I’ve gained over the years I’ve condensed the content into 10 rules I apply when I conduct research that prove to be useful: