A typical workflow of AfL is shown in the diagram below.
- We assess where the learner is now through the use of formative assessment tasks
- We find out where the student gaps are
- We find ways to address them
With these three parts of the workflow, we can potentially see where AI can help us.
For us to determine where the learner is, we will need to provide them with questions to evaluate their understanding. After we provide them with feedback, we can provide them with opportunities to work on the feedback, before providing a post-test to ascertain if they have understood the feedback accurately. This means that the creation of questions of similar nature is important. Perhaps where AI can play a part here is to assist teachers in the creation of similar questions, so that the teacher can have a good set of questions to test their students, both pre and post-test. AI tools such as ChatGPT can come in to create such questions.
On top of creating questions as prep work, what about in-class formative assessments? As of now, teachers can provide students with in-class assignments, grade the work, and return them to students. As mentioned earlier, this could be rather time-consuming. However, quiz tools such as Kahoot and Quizizz allow teachers to run in-class multiple-choice questions (MCQ) and get immediate feedback. This has proved to be effective and hence is part of many teachers’ toolkit of EdTech tools to use. Despite that, MCQs do not capture as rich information as short-answer texts. However, the issue with capturing short-answer texts is that teachers do not have enough time to read through all responses quickly — again a scaling issue.
Hence, what if there is an AI tool that can be used in class that allows for the clustering of student responses, perhaps it might help teachers go through multiple responses in a shorter time. The two pictures below show a sneak peek of the tool that I am working on, which allows students to submit their answers in real-time, and the program will auto-grade and cluster student responses.
Beyond that, adaptive learning systems (ALS) and AI tutors can also be promising avenues for us to capture information about student learning.