AI Tools for Scholarly Research: GenAI Research Ethics

When we discuss generative AI (GenAI), we are most likely discussing large language models (LLMs) that are able to generate responses based on their large training datasets. Due to the nature of this training and how these programs operate, there are a number of ethical concerns when using them for scholarly or non-scholarly research.

These include:

• Disparity in free versions vs paid versions of these tools

  • Those who are able to afford to pay for these tools will (generally) have access to more features and more effective searching.

• Training data collection

  •  Some of the data used to train LLMs was the intellectual property of others and used without permission. This is problematic from both a legal standpoint and an ethical one since it's not always clear where information for the answer to your question comes from.

• Ability to cite information and determining sources/authority

  • A vital part of research is citing where your information comes from to back up any claims made in your paper or project. Answers from LLMs can sometimes not include this information and you cannot verify the veracity of the information. This is a reason to use LLMs for brainstorming and writing review and not for information gathering.

• Accessibility

  • Not all GenAI tools are fully accessible to all- there are potential financial and accessibility issues for those who use screen readers or use other adaptive technology.

• Use of GenAI vs personal intellectual creation

  • Some users may prompt a tool to do their assignment for them. This not only robs themselves of intellectual growth but could also be a violation of student conduct rules.

• Environmental concerns

  • A large amount of energy is used each time a GenAI tool is prompted for information. There are some ideas on how to reduce or mitigate this issue but currently this is a significant issue as more individuals begin using this technology.

Here are ways to address these ethical concerns:

• Use GenAI products as another tool for your research

  • Instead of asking a large language model like ChatGPT to write your paper or part of it, use it to help you brainstorm ideas, develop search terms to use in library databases, plan your project out, etc.

• Cite your GenAI source

  • If you do use a GenAI tool to create something (e.g. to visualize data), properly cite the use of it in your work.

• Minimize use for environmental impact

  • Once you determine your project outline, topic(s), and search terms, use library resources or Google Scholar to minimize the use of energy (compared to GenAI tools).

• Be careful utilizing GenAI for information gathering

  • Due to the training data used for large language models, asking these models factual questions can be problematic since you cannot always identify where the information came from and if it was an authoritative source. If sources are provided, try to find other sources of trust information that can back up that information.