Top 5 Risks of Using AI for Academic Referencing Purposes

While many PhD students leave referencing for last-minute draft revisions, this element forms the core of academic research. The quality, recency, and relevance of your sources directly influence the novelty of your study. The integrity of your references also demonstrates your knowledge of the field, intellectual lineage of your findings, and your overall respect for and recognition of existing theories and debates in your field.

AI-generated reference lists look like a good idea at first. You get a professional-looking collection of articles with ready-made summaries you can instantly cite in your literature review. However, this convenience can be much more dangerous than you expect. In this article, we will analyse the top 5 risks of using AI for academic referencing purposes.

1. Lack of Full Control over AI Choices

Most AI platforms look like a black box even to professional IT specialists. While some can be asked to explain their reasoning explicitly, the motivation for choosing particular papers in response to your request remains unclear most of the time.

• Are these articles the most recent ones in the training database?
• Were they prioritised according to some undisclosed central tendency?
• Are they the easiest to gain access to (i.e., older)?

Explaining your choice of sources for a Bachelors or Masters dissertation can become difficult if you suddenly miss some recent breakthroughs in your field for no particular reason. This may become an even greater problem in PhD research, where the selection of sources has to follow a strict and explainable logic.

2. Amplification of the ‘Mainstream View’

As noted above, AI frequently follows some central tendency when answering requests. In academic research, this means that some points of view can be overrepresented. This may occur with specific regions of the world, theories, and seminal works of emerging scholars. The ‘academic cost’ of such selection is the narrowness of analysis and potential loss of important literature in your field. Also, AI cannot separate primary and secondary sources or trace back some ideas to their original authors. This frequently leads to one-dimensional analysis missing criticism and overemphasising some standpoints that were more widespread in the data the model was trained on.

3. Fabricated Sources

While popular AIs can get some basic facts right most of the time, accurate referencing remains their Achilles’ heel. According to recent articles, from 18 to 69 per cent of citations can be fabricated depending on a particular model and research area. The worst thing is that AIs do not actually tell you which sources are the result of their hallucination. This leaves you with a reference list you cannot fully depend on. In many cases, the journal names, authors, and article names look highly realistic since the algorithm combines pieces of real data to create fake items. If you are not planning to spend sufficient time checking every reference on your list, you may reconsider the idea of using AI for referencing purposes.

4. Inaccurate Summaries and Lists

In many cases, AIs are used by researchers to generate lists of key literature in a particular field and summarise the key points of recent publications in it. On paper, this looks convenient for developing your literature review and saving a lot of time. Unfortunately, these features are still too unreliable to guarantee good credibility. AIs mix primary and secondary sources and do not verify the quality of specific sources. This means that your summary can include studies concealing their methodologies, non-peer-reviewed articles, and even AI-generated articles cited in real articles. The summaries also lack criticism since the algorithm simply ‘retells’ the contents instead of approaching the data critically and positioning it against a wider network of knowledge in this sphere.

5. Mix of Functions

If you consider the bigger picture, citations and references serve several functions. On the one hand, they address instrumental and professional aspects by showing the depth of your research, your understanding of key debates, and your contribution to the field. On the other hand, they also serve a rhetorical function by helping you support your claims, contextualise your arguments, and position your findings within existing studies. While ideal AI systems can potentially address the first elements, they cannot help you develop your own ideas or create them for you. Moreover, they largely guide you towards specific arguments instead of facilitating your independent thought. As a result, many students start using them as a crutch rather than a support, which leads them to generic findings informed by available sources instead of novel research going beyond what we already know.

While the lure of using AI systems for findings and summarising your key references is extremely strong at the moment, you should always be aware of the risks above. These platforms can give students a wrong sense of confidence that they promptly lose when they encounter the problems stemming from fake sources or incorrect summaries of key ideas. If you find yourself confused by the task at hand, try seeking professional writing help to get expert guidance on referencing and finding the best sources to support your research.

If you are struggling to find sufficient sources for your PhD literature review or need help compiling your final bibliography before submission, our PhD writers can help. Outsource the longest task of your PhD dissertation writing, like PhD editing, to the experts at Original PhD.