What is GenAI?

GenAI is the sub-area of AI involving systems which ‘generate’ content on a user’s behalf. Where data analytics and other forms of AI previously focused on analysing text, images, and speech, GenAI generates new content based on the user’s question, query, instruction or prompt tailored to the user’s needs, requirements or instructions. The content it generates can take the form of explanations, plans, process descriptions, questions, images or conversational dialogue. However, GenAI tools do not always generate fully correct answers and those using it are advised to read about the risks and restrictions on its usage.

GenAI has been defined as a 'game changer' for society (World Economic Forum 2023) with significant implications for higher education. As a result, it is essential that we understand what GenAI is, how it works, and how GenAI can be used ethically and responsibly to support teaching, learning, assessment and research — for which developing AI Literacy has become essential.

How do I use GenAI?

There are many GenAI tools available online (e.g. ChatGPT, Microsoft Copilot etc.) which are accessed via a query/conversational interface. These tools typically ask for a ‘prompt’ (in the form of a question or instruction) typed into a text space. To start using GenAI is thus very easy. If you can phrase a query or a question, you can use GenAI.

GenAI tools can generate very eloquent, convincing text and images. However, GenAI tools do not store facts and knowledge, rather they generate outputs based on probabilities. Thus, GenAI is prone to making factual errors (called hallucinations) which are nonetheless very convincingly presented. This is where academic/professional judgement and domain expertise are very important. If you are using GenAI, you need to double check the information it is giving you as it will present information which is simply not true!  Therefore, it is crucial for any user to fact-check any output from a GenAI tool.

GenAI has proven to be a very effective tool for exploring information, suggesting activities or plans appropriate to a problem or task, or generating ideas or materials through interactive dialogue.

How do I cite usage of GenAI?

Where the output of GenAI is used in a document or work output, this usage should be acknowledged and appropriately cited. A citation should typically include the date of generation, tool used and prompts used to create the output with verbatim quotations enclosed in quotation marks. School/Degree programme handbooks should provide rules for such citation. The format of the citation is dependent on the type of work output or document for which it is being included. Where GenAI content is used verbatim (e.g., in the form of unedited text or image), this should be accompanied by a full citation, with text-based content included in quotation marks. For further guidelines on this see Guide to acknowledging the use of generative AI and referencing generative AI (developed by the Library of Trinity College Dublin).

Note any long verbatim quotation (e.g. more than one paragraph), even with citation, may be considered inappropriate or poor practice in student assessment documents and publications.

What should you NOT do with GenAI?

It is important to understand that some usages of GenAI are unlawful and must therefore be avoided.

Many GenAI tools are trained on vast amounts of data gleaned from a wide variety of sources. However, the training of such tools is not transparent and the exact extent of their training data and sources remains unknown. Some tools that have been trained on material on the open web are likely to have ingested protected personal data, copyright-protected content, copyright-infringing content, misinformation, disinformation, hate speech, defamation, and all manner of other unlawful content. Such models are then likely to produce unlawful material in their outputs. However, even GenAI tools trained on curated data to avoid such illegal inputs can still be used in unlawful ways and can still produce unlawful outputs –  and it is necessary to be aware of these problems when using such tools.

For example, since most GenAI tools harvest inputs and use interactions with users for their systems development, you must ensure that their inputs, prompts, queries, instructions, contextual information, and other interactions are lawful. So, just as you are not allowed share personal, private or sensitive information about colleagues/students on websites or via other electronic means, you are NOT allowed to use such information as part of inputs, prompts, queries, instructions and other interactions when using GenAI tools. In most cases, to do so is unlawful as you are sharing private information with a third party (the GenAI provider). Even where sharing such private information is not necessarily unlawful, it is against College regulations. Hence, student work (submitted assessments and contributions) are considered private information, and are NOT allowed to be uploaded into a third-party GenAI tool for any reason.

Similarly, content which is confidential in Trinity or confidential to your user’s studies or work (research, teaching or administrative) or for which you do not own the copyright, or which is not publicly available, should NOT be used in creating inputs, prompts, queries, instructions, contextual information, and other interactions for GenAI. Again, in most cases, to do so is unlawful. Even where sharing such information is not necessarily unlawful, it is against College regulations. Hence, confidential College information is NOT allowed to be uploaded into a third-party GenAI tool for any reason.

What are the key concerns with using GenAI?

As well as the legal liability concerns mentioned above, there are several other concerns of which users of GenAI need to be aware. Because GenAI is trained on such a wide pool of data, content generated by a GenAI tool can contain factual errors and exhibit bias (which can come from bias already embedded in its training data).

The training and use of GenAI systems can also use significant amounts of energy and resources, leading to sustainability concerns. This energy consumption should be considered in relation to College Sustainability policies and practices. Additionally, some GenAI tools harvest information from user prompts including contextual information from users’ interaction with GenAI tools, leading to privacy and intellectual property concerns. GenAI tools vary in regard to the extent of these concerns.

How can you use GenAI in your research?

Knowledge and understanding of how GenAI tools can be used for research is developing rapidly. Many of the issues and responsibilities covered in previous answers are highly relevant regarding lawful, ethical and sustainable use of GenAI in research. The accuracy of AI-generated content needs also to be carefully considered and biases mitigated. Citation of results from GenAI usage in published papers (conference and journal) are determined by the publisher/event organisers/professional body and adherence to their guidance needs to be maintained.

Regarding how GenAI use can be embedded into the research process and activities, many possible approaches are being explored e.g. using GenAI to correlate papers in literature reviews, using GenAI to generate possible approaches to a problem or task, using GenAI to iterate through and exhaustively evaluate possible solutions/approaches etc. Usage of such tools should be consistent with Trinity’s Policy on Good Research Practice.

What advice should I give students on their use of GenAI?

GenAI will be used by students, researchers and staff. Given the ubiquity of GenAI tools within everyday devices and platforms, it is impossible to ban their use. Students and staff are advised that GenAI can generate erroneous, biased content. Therefore, students and staff are expected to fact-check information generated by GenAI, and to seek out primary sources of information (e.g. reputable books, publications, papers etc.) as part of a rigorous academic practice. Further advice is available on the Centre for Academic Practice’s GenAI Hub and the National Academic Integrity Network’s Generative AI Guidelines for Educators (July 2023). 

From an academic integrity perspective, if a student generates content from a GenAI tool and submits it as his/her/their own work, it is considered plagiarism, which is defined as academic misconduct in accordance with College Academic Integrity Policy. If a sentence or quotation from GenAI content is  used by a student in their academic work, it must be referenced. Cases of plagiarism are considered under College Academic Misconduct Procedures and College’s Academic Integrity Policy. A complete guide to Academic Integrity policies and procedures is available on Academic Affairs’ Academic Integrity Policy and Related Procedures page. Many research funders and publishers now have policies in relation to the use or misuse of GenAI in writing and review for publications, funding proposals and other academic outputs. Researchers are advised to check any restrictions set by funders and publishers on the use of GenAI.

The Library of Trinity College Dublin has also developed guidelines on acknowledging and referencing GenAI. A link to these can be added to your Blackboard module and/or course handbook.