Education, Organisation, Research
Using AI in written assignments: four levels for lecturers and students
Leiden Law School’s AI Working Group has formulated a clear framework within which generative AI (Gen AI) can be used in written assignments (including models by OpenAI, Anthropic, or Google).
Menu of options
The framework is designed as a menu of options which lecturers can consult to decide:
- Whether the use of GenAI is allowed;
- what students are allowed to use it for;
- why that choice was made;
- the opportunities and threats students need to be aware of; and
- what students will be assessed on.
It provides lecturers with practical guidance to align the use of GenAI with the learning objectives and the students’ year of study. This helps prevent students from outsourcing everything to GenAI and learning nothing, or blindly trusting GenAI output without checking and correcting it.
Implementation
The Faculty Board has approved this framework, and plans to implement it in the education policy of the faculty. The framework will be included in the Prospectus for the new academic year. Teaching staff therefore have ample time to decide on the role of GenAI in their written assignments, and students will know what to expect. The framework will also ensure a consistent approach that is recognisable to both students and lecturers.
GenAI level in line with learning objectives
Tycho de Graaf helped to create the framework: ‘Lecturers can use the table to decide what students can use Gen AI for. You do that, of course, depending on what you want to teach them. To help, the table specifies for each GenAI level how our standard learning objectives and assessment criteria should be applied in view of the permitted use. Think of Word’s spelling check: we evaluate spelling knowing that students use that tool and that we allow it.’
Faculty Talk and AI Symposium
On Tuesday 26 May, Tycho de Graaf will give a Faculty Talk on GenAI in education in the run up to the faculty Symposium: ‘Moving forward together with AI’ which will be held on Thursday 28 May. The symposium will focus on GenAI in research, education and the organisation and is intended for all staff at our faculty. Please register here.
Using GenAI in written assignments
A brief outline of the table is provided below. For more detailed information, please go to the LTC HUB.
Level 1: prohibit the use of AI
Students learn to independently develop research questions, analyse sources, create structures, construct arguments, and compose text. This level encourages critical thinking and self-reliance which are both essential for effectively assessing AI-generated output in the future.
Tycho de Graaf: ‘The lecturer explains the rationale behind their decision. This also underscores the ongoing importance of learning how to conduct research and write without GenAI, which will help students better check the accuracy of GenAI output when they are allowed to use it.’
Level 2: allow its use to edit text
GenAI may be used to improve text, for example spelling, grammar and source references. There are two options: Using GenAI during the entire writing process or only for the final version. Students remain responsible for the content and learn how to critically assess output from AI.
Tycho de Graaf: ‘If the goal is to teach students to use GenAI to improve text they have written themselves, then the student must explain how and which GenAI tool was used in the assignment. This lets the lecturer check whether GenAI was used in a good way. Lecturers assess whether the text is focused on the topic and meaningful, and look beyond just error-free language and smooth sentences.’
Level 3: allow or require its use for brainstorming, source research, and creating initial structures
GenAI is allowed or required to come up with research questions, find sources and create an initial structure. Students are not allowed to use it to generate text. They learn to use GenAI as a tool but remain responsible for the final content and argumentation.
Level 4: allow or require its use for performing certain tasks or all tasks
GenAI is allowed or required for tasks such as composing text or constructing arguments. Critical reflection and human evaluation remain essential. Students learn to use GenAI, and to assess, adapt, and account for the output.