You are here

Generative AI and Authentic Assessments

Generative AI technologies are effective tools that will generate plausible sounding text on any topic and compose answers to ranges of prompts and problems, such as writing essays, solving math problems, developing HTML code, etc. 

If a course contains assessments that require students to perform these same tasks, students may be tempted to use Generative AI tools to help them complete their work.

Many of these tasks make up authentic assessments, which have been promoted as a way for students to “authentically” demonstrate their knowledge and skills in a more open-ended and real-world manner than rigid traditional assessments. They allow students to be more engaged and involved with the subject matter and can encourage participation from those with different learning styles and interests. 

Some characteristics of authentic assessments include the following:

  • They can be research papers, projects, presentations, demonstrations, simulations, etc.
  • Students work outside of a supervised environment over a long period of time.
  • Collaborative work is often included.
  • Students have access to technology, including generative AI, throughout their progress.

In contrast, the hallmarks of more traditional assessments include the following:

  • Feature the use of closed-ended questions, such as multiple-choice, matching, true-false, etc.
  • Essay assignments are shorter and rely on timed responses.
  • Testing occurs in a heavily controlled, supervised, timed environment.
  • Students do not have unplanned access to technology, including generative AI.

Due to the heavily controlled and supervised environments in which traditional assessments are delivered, they tend to be less affected by the new generative AI tools. Those using authentic assessments, on the other hand, find themselves needing to re-examine their courses to see how AI might affect their current assessments.

There are two major ways to respond to the presence of generative AI technology when using authentic assessments: prevent students from using AI or encourage students to use AI.

Preventing AI Usage

Preventing students from accessing Generative AI tools in an authentic assessment is very difficult. AI detection software has proven to be unreliable. Some educators are choosing to scale-down their authentic assessments to a more traditional, monitored format. 

Assessments that focus on student strengths and/or AI weaknesses may discourage students from relying on AI. Some examples include:

  • Ask students to share their personal feelings or experiences.
  • Include an oral or multimedia presentation with a Q&A segment.
  • Create assessments with smaller components that work together to form a final project.
  • Consider projects that involve data collection or experimentation.
  • Add assessment components based on a classroom discussion or a specific context unavailable to AI.

Encouraging AI Usage

Free, robust Generative AI tools are currently being used in real-world professional settings. Therefore, it may be helpful to prepare students to use these tools effectively to think with them and not think for them.  Additionally, it may be physically impossible to prevent students from accessing this technology. Educators can choose to embrace this change and ask students to perform tasks where generative AI use is part of the process.

Examples of assessments that require students to use AI tools include:

  • Ask students to craft a prompt that will generate a specific output.
  • Have students analyze an AI output for factual errors or fictional responses invented by the AI (hallucinations).
  • Ask students to compare the outputs of different AI tools.
  • Have students make critical improvements to an AI-generated outputs and document their work.

Keep Students Informed

Whichever approach you decide, keeping students informed of your expectations is crucial. Consider adding a syllabus statement that clearly details if, how, and when the use of AI technology is appropriate in your course.

Sources:

Authentic and Traditional Assessments

Generative AI and Assessments