University educators have a duty to prepare students for an AI-enabled world. This can be challenging, given the plethora of tools available; the temptation to try every shiny new one is common, but doing so without careful consideration may raise ethical concerns.
So which AI tools should we use? And how should we incorporate them into our teaching, learning and assessment sustainably? To understand this, we must build our own AI knowledge first.
What’s already approved?
The first step is to find out which (if any) have been approved by your institution. IT staff, together with the data protection team, may have reviewed tools for general use, and these would be ideal starting points for experimentation. For example, Microsoft or Google institutions will likely find that Copilot or Gemini are approved and secure to use. This means that data you enter into these tools are held securely within the “digital walls” of your institution. At my university, our approved tools are Copilot, TeacherMatic, Adobe Firefly, and careers-focused tools CareerSet and Shortlist.me.
We recognised the importance of tool governance early on in our AI journey. Our approach is informed by our guiding principle to cautiously, transparently and responsibly embrace AI by balancing the benefits of use with the need to evaluate tool security and data protection. The committee that reviews tools includes representatives from information security, data protection, pedagogy and other specialist areas.
What happens if the AI tool I want to use isn’t on the approved list?
In many cases, there will be tools that are appropriate to your course but are not on your institution’s approved list. We use the SafeAI framework, which we adapted from Charles Sturt University’s S.E.C.U.R.E. framework, to provide a way for colleagues to review tools they would like to use for use in their disciplinary contexts. It uses a questionnaire to guide them through whether the tool can be used within the spirit of our principles. It does this by prompting the user to consider whether they would be entering personally identifiable, confidential, security-related (passwords and usernames) or copyrighted information into the tool.
It also asks users to reflect on whether they are using the tool to perform unethical activities, and whether they would be using the outputs without evaluating them. If the user answers “no” to all the questions, they would then be able to use it for the specified context.
Cost is a key factor to take into account when choosing an AI tool. When one is offered for free it is likely that your data you enter is the “cost”. All major chat-based tools have settings that allow users to opt out of their data being used to train the model.
Some tools are “better” than others – not necessarily in terms of their outputs, but in how they handle and manage data or security. End-user licence agreements (the legal contract between a user and a software supplier) and privacy policies are usually available on the respective tool’s website, so check these before use.
For example, you might choose to use tools like CareerSet that store their data in the EU, which would require the supplier to adhere to robust data protection under the General Data Protection Regulation (GDPR). Or you might choose tools like Microsoft 365 Copilot for Enterprise that don’t use user data to train their models, or tools like TeacherMatic that have a time limit on storing input/output data, and multifactor authentication or other security measure.
Consideration of the data you intend to use within the AI tool is also important. S.E.C.U.R.E. highlights the importance of protecting personally identifiable information and considering intellectual property – for example, it might contravene the terms of use for you to enter a paywalled journal article into a tool to create a summary.
Embedding tools in your teaching: career-readiness and conscientious objection
Some students may object to using AI either on sustainability, creativity or ethical grounds. While the reasons for AI objection are often deeply personal, it is important that we uphold our duty to provide guidance on the ethical and transparent use of AI. This also involves modelling ethical and transparent use.
AI is being embedded into new and existing tools and processes at pace. To prepare our students for graduate roles, we need to understand how the tools we use and teach our students operate, what the risks are and whether they are fit for purpose. Then we can begin to embed AI literacy into our teaching and learning processes cautiously, transparently and responsibly.
Laura Milne is the head of digital education at the University of Chester.
If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.