Generative AI and library skills

In considering Generative AI (GenAI) as a source of information, we are normally referring to Large Language Models (LLMs). LLMs are GenAI systems that generate text by statistically analysing the distribution of words in the data they have ingested and identify and repeat common patterns (for example, which words typically follow which other words).

Generative AI cannot be seen as an original source of information, and the information it generates cannot be replicated or referred back to by others.

(Image generated by Open AI's ChatGPT using Microsoft CoPilot)

GenAI should not be used in isolation as a source of information. You should always verify information from GenAI and use it alongside reliable search tools, such as UCL Explore and library databases. Your subject guide recommends sources of information for your discipline.

GenAI has limitations as a source of information, and should be used with caution. If you choose to use GenAI as a source of information, it can be particularly helpful to:

• help scope and focus your research question, generate and formulate your ideas.
• get background information to inform your topic or get definitions of unfamiliar terms.
• provide summaries of complex information to help you understand a topic.
• translate text into your own language. Note that GenAI can make errors in translation, and you should check the permitted use of machine translation tools in the context in which you are using it.

As GenAI outputs may be inaccurate, biased or out of date, you should always follow up by verifying the information from the original source and / or other credible sources. GenAI can assist in informing your search for information using credible sources. See GenAI-assisted searching.

Note: These uses may require you to use specific and considered prompts to achieve effective results. See Effective prompting.

• GenAI is not an original source of information; it generates information from other, often unidentified sources.
• GenAI does not understand the content it generates, which can lead to hallucinations or factual errors, or information that may be biased, offensive or outdated.
• Unless GenAI tools are integrated with an internet search engine (e.g. Copilot) or incorporate external data sources via a plug-in, GenAI tools do not have access to live data and are dependent on the data on which they were trained.
• GenAI may fabricate quotations and citations.

General Large Language Models (LLMs), such as Microsoft Copilot or ChatGPT, can be a useful place to start exploring your topic or scoping your research question. 

If you choose to use a general LLM, we recommend you use Microsoft Copilot. UCL staff and students can access Microsoft Copilot, which can be used for both text and image generation. With commercial data protection, this is intended as a more secure alternative than other GenAI services. If you wish to use GenAI, then this is the safest way to do so. 

If you're logged into Microsoft Copilot with your UCL credentials, what goes in – and what comes out – is not saved or shared, and your data is not used to train the models.

In addition to broad limitations of GenAI, an additional issue with using a general GenAI tool as a source of information is that the output will not always include links to the original sources of the information, and where it does, you will often find the sources are unreliable or non-scholarly sources. You may therefore prefer to use GenAI tools specifically designed as sources of information for academic research which draw on reliable sources. Find out more on our page on GenAI search tools.

If you choose to use GenAI as a source of information, always ensure:

• you are permitted to use GenAI for your assignment or research;
• you understand the limitations and risks of using AI as a source of information;
• your assignment or research remains your own work;
• you critically evaluate output from GenAI and check information against original sources;
• you document your use of GenAI so you can acknowledge your use of AI appropriately.