4 tips for creating a successful AI assignment
Lulu Gao | firstname.lastname@example.org
Jan 5, 2024
Illustrated by AI, written by humans.
What is an AI assignment?
What makes an AI assignment “successful”?
A "successful" assignment goes beyond helping students memorize. It’s a step forward in revolutionizing your teaching methodology, not just substituting parts of what you already do. Some questions you could ask to reflect on this include:
Are students mimicking what they’ve seen before, or are they reasoning solutions for themselves?
Are students engaged and actively thinking, or are they learning passively?
Are students intrinsically motivated, or are they just striving for the bare minimum as defined by assessment standards?
Each teacher’s exact definition of success will be different, but these are general reflections we suggest having before determining how you want to use AI with your class. Now, let’s cover more specifics on creating your AI assignments.
Tip #1: Setting the context
AI models have a vast wealth of knowledge to pull from. At first, this might lead you to assume that you won’t need to add resources or context to your assignments. However, uploading relevant content can give the AI a better idea of what you want it to focus on in the assignment. For example, you can ask the AI to:
Cover only certain parts of a book without spoiling what’s to come by uploading specific chapters.
Cite a primary resource in its conversation about a larger topic by attaching the resource link.
Practice specific words and phrases with students by uploading a vocabulary list.
It’s also important to give the AI a concise and pointed learning objective. Explain what you want in detail, but not too much detail because the AI doesn’t always know how to prioritize the different phrases in your request. Having too detailed or too long of an objective can blur the main focus of the assignment and limit the creativity of the AI’s output. It’s better to start broad and make revisions to hone in on how you specifically want the AI to behave.
Tip #2: Preview and iterate
Revision is key to crafting the specific learning experience you want. Being able to trial and preview the AI’s expectations and assessment will help you spot strange or unwanted behavior. Preview good, mediocre, and poor student responses to make sure they fit your expectations of how your students should respond and how the AI should guide and assess them. Additionally, use the manual preview method to ask any specific questions you anticipate students having. You can also check how the AI will respond to any specific situations you’re concerned about. As you find mistakes, you can use the revise feature in Flint to ask for edits in plain language and Flint will make sure the revisions happen across all inputs.
Remember, designing AI is not a game of perfection. Embrace the weirdness! Part of helping students learn AI literacy is showing them how AI can make errors and leveraging those moments as learning opportunities.
Tip #3: Don’t be afraid to break the mold!
AI has illuminated new frontiers for education, promising the ability to tackle long-standing problems of practice and bring about new possibilities for pedagogy. This is why we built Flint to be flexible and fit teachers’ aspirations without any set curriculums or hyper-specific tools. Our advice here would be to work with other teachers to develop AI strategies: learn from your colleagues, share your own findings, and collaborate across classrooms.
Some creative ways we’ve seen teachers use Flint in and outside their classrooms include:
Creating untimed assignments to serve as general-purpose tutors that students can access at any time.
Projecting Flint on the board and using it to guide a class-wide discussion, making a lecture more dynamic and participatory.
Having the AI guide collaborative, in-person group work—similar to how a teaching assistant or subject matter expert might interact with curious students by asking questions, challenging assumptions, prompting further thinking, etc.
In case you’re curious about how AI can be used for a particular subject you teach, you can check out our use cases page. There, we’ve compiled subject-specific examples and testimonials from teachers.
Tip #4: Think beyond the class period
Remember, there is no such thing as a perfect AI assignment—just like there isn’t one best way to teach a unit, there isn’t one best way to incorporate AI into your lessons. Seeing each AI assignment as a learning opportunity that informs the next assignment is important for both you and your students to learn the abilities and limitations of this technology.
We suggest bringing your students into the equation and letting them have a say in how AI should impact their learning. This gives them both a sense of agency and investment in the evolution of their education. We’ve seen Flint teachers survey students about what they liked and disliked about the AI assignments they completed. One of the surveys showed that students loved the AI’s personalized feedback but didn’t like the timer cutting their conversations short. Students also directly said they would want more guidance and practice with how to ask questions to the AI. This helped the teacher learn what types of assignments to give in the future and what additional instruction might be needed to promote student AI literacy.
Ultimately, the students are the ones who will be using AI for learning, so involve them in the process early on and determine what inspires and motivates them. This can better inform what you as a teacher can do to prepare your students to shape their own futures.