How Do I (Re)design Assignments and Assessments in an AI-Impacted World?

How Do I (Re)design Assignments and Assessments in an AI-Impacted World?

​ Group of three diverse professors who wonder about assessment of student learning in an AI driven world, AI style.This image was created using Bing Creator with the prompt: Group of three diverse professors who wonder about assessment of student learning in an AI driven world, AI style. [Click and drag to move] ​
Caption: Bing Creator Prompt: Group of three diverse professors who wonder about assessment of student learning in an AI driven world, AI style.

In an era where AI tools are readily accessible, educators may want to consider transforming their approach to designing course assignments and assessments. Generative AI tools can be used by students to complete out-of-class assignments with or without teachers’ knowledge; however, the use of AI output detectors has raised concerns about their accuracy and inherent biases (Trust, 2023a) as well as concerns as to how their use can impact instructor-student relationships. Torrey Trust, UMass Amherst Professor in Teacher Education and Curriculum Studies, emphasizes the need for redesigning assignments to reduce the potential for students’ inappropriate use of generative AI tools (Trust, 2023a) and advocates for making the purpose and requirements for assignments as transparent as possible, increasing the value of assignments through real world applications, and emphasizing learning as a social experience through assignment design (Trust, 2023b).

Rethinking and transforming assignments and assessments offers a potential opportunity to enrich students' educational experiences. It may be helpful for faculty to approach this transformation by weighing strategies which assignments or assessments should or could be AI-proof or AI-immune, integrating AI tools as part of learning activities, or by adopting a two-lane approach which combines these strategies (Liu & Bridgeman, 2023).  

Strategies and Examples

General Considerations

Align Objectives. Identify course goals and student learning objectives and reflect on the objectives to determine which ones require students to demonstrate learning in an AI-free environment or through AI-immune assessments, and which new objectives related to AI skills and knowledge you might want to add to your course. Consider alignment with AI-integrated assignments or a two-lane approach' where appropriate.

Communicate with Students. Be transparent with your students about your rationale for your assignment design choices as well as your expectations and parameters for the use of AI tools for each assessment or assessment category. Talk with students about the affordances, limitations, and risks with using generative AI tools, and emphasize that they should never input any sensitive and/or personally identifiable information. See the CTL webpage How Do I Consider the Impact of AI Tools in My Courses?  

Minimize Learning Barriers. Be mindful of the potential to increase learning barriers with low-tech options, such as requiring handwritten essays or oral in-class presentations, and the unreliability of AI text detectors (Trust, 2023a) as well as equity issues related to student technology access.

AI-Immune Assignments

Designing assignments for critical thinking, personal reflection, or unique insights that AI cannot replicate ensures authentic learning. Examples include:

  • In-Class Presentation with Q & A
  • Gallery Walk Activity: Students explore various stations featuring different topics, questions, or projects, engaging in discussion and reflection.
  • Analysis of unexplored real-world scenarios in class: After students analyze and discuss their scenarios, ask them to visually represent their findings on posters.
  • Class Discussion Synthesis: Require students to integrate insights gained from in-class discussions into their writing.
  • Complex Multilayered Project: Assignment has multiple interrelated questions or tasks that build on each other, and students will need to produce highly detailed responses that meet high field-specific standards and/or language and they need to cite specific passages of the content informing their project (Clay, 2023).
  • Source Documentation: Require students to screenshot the relevant passages that they reference in a paper or include annotated versions of their sources with their submission.
  • Process Documentation: Require students to turn in submissions in Microsoft Word or Google Docs with the version history turned on.

Read about putting assignments to the test in this AI-Immunity Challenge.

AI-Integrated Assignments

This approach involves purposefully incorporating AI technologies into assignments for assistance with specific tasks. Begin by identifying learning objectives that explicitly involve using AI tools, such as the ability to analyze data using AI tools or the ability to use AI-generated feedback to improve one’s writing. UMass IDEAS provides an excellent overview of various Generative AI tools and the resources available on campus. Consider using the TILT Transparent Assignment Template to provide your students with a clear understanding of the purpose, expectations, and parameters for an AI-integrated assignment. Here are some examples:

  • AI-Assisted Research Project: Leverage AI for various research tasks.
  • Document AI Collaboration: Students use specific AI tools, documenting and reflecting on the process.
  • Judge the AI Output: Guide students to evaluate AI-produced materials critically.
  • Chart Your AI Journey: Assign students a performance task, such as designing a product, creating a plan, preparing for a structured debate, creating an art piece, or creating a multimedia presentation. Instruct them to research, select and utilize relevant AI tools to complete it. They should then outline and justify their chosen approach.
  • Chatbot Conversation Analysis: Analyze chatbot responses, effectiveness, and biases.
  • AI Art Critique: Compare AI-generated art to human-created works.
  • Ethical Implications Debate: Engage in discussions on AI's ethical considerations.
  • Data Visualization Project: Guide students to use AI tools for analyzing large datasets, creating visuals of findings, and discussing patterns, anomalies, and insights.

Be mindful that Copilot (formerly know as Bing Chat) is included in the UMass agreement with Microsoft; it is considered FERPA compliant as long as users have logged in using their UMass accounts. If you are considering requiring students to use an AI platform to complete coursework, Copilot is the only recommended option. Note that with any online service that is not contracted by the university, in view of FERPA regulations, students must not be required to identify themselves to third parties. If you plan to AI-integrate assignments using non-supported tools, provide an option for students who wish to complete the assignment without using AI tools.  

A Two-Lane Approach: Combining AI-Proof and AI-Integrated Assignments

Design some tasks that cannot be completed using AI tools, for example by having students complete the assessment in an AI-free environment or designing an AI-immune assignments (lane 1) and others that involve AI collaboration (lane 2). By adopting this two-lane approach (Liu & Bridgeman, 2023), educators can leverage AI to create a balanced and enriched learning environment that both safeguards against potential AI misuse and leverages AI's unique capabilities to foster creativity and deeper understanding.

Liu (2023) includes several examples of integrated AI assignments and the two-lane approach in this guide: Assessment Design Guide by Danny Liu, University of Sydney 

Consider

The redesign of assignments and assessments for an AI-impacted world is not just feasible but also an exciting opportunity to enrich the students’ learning experiences. By understanding how AI may alter learning objectives and creatively integrating or safeguarding against AI tools, you can create a robust educational environment and help students to meet new and evolving demands in their future workplaces.

 

Resources

Essential Considerations for Addressing the Possibility of AI-Driven Cheating, Part 1 - Torrey Trust, College of Education, in Faculty Focus August 2, 2023 (Trust, 2023a)

Essential Considerations for Addressing the Possibility of AI-Driven Cheating, Part 2 - Torrey Trust, College of Education, in Faculty Focus, August 4, 2023 (Trust, 2023b)

What to do about assessments if we can’t out-design or out-run AI? - Danny Liu and Adam Bridgeman, July 12, 2023

Assessment Design Guide by Danny Liu, University of Sydney 

Embracing Constructive Dialogue and Oral Assessments in the Age of AI-Graham Clay and Cambriae W. Lee in Inside Higher Ed August 3, 2023

How Do I Consider the Impact of AI Tools like ChatGPT in My Courses? | Center for Teaching & Learning (umass.edu)

A Guide: How Professors Can Discourage and Prevent AI Misuse - Graham Clay, August 7, 2023 (Clay, 2023)

Infographic Generative AI Tools List

Searchable AI Tools List by Future Tools

Incorporating AI in Learning Assessments: A Guided Pathway - Sean McMinn, June 15, 2023 (LinkedIn Blog Post)