8 Ways to Use Generative AI for Student Assessment

Finding ways to assess student learning outcomes in a manner that is accurate, effective, and engaging has always been one of our biggest challenges—and opportunities—as educators. As technology has evolved over the years, we have all had to evolve with it. Today we find ourselves facing ChatGPT, the generative AI tool launched in November 2022 that caused equal parts panic, curiosity, fear, and fascination among the public—and these emotions have been felt vehemently in education.

While artificial intelligence (AI) itself is nothing new (the technology behind it has been around since 1950 thanks to mathematicians Alan Turing and John Von Neumann), nor is the machine learning that generative AI is based on (also around since the 1950s thanks to Donald Hebb, Arthur Samuel, and others), their accessibility to the general public and the rate at which the technologies have been evolving has never been greater.

With ChatGPT boasting more than 100 million users and garnering nearly two billion site visits per month, there’s no question as to whether students are using it, but rather how they are using it. Generative AI tools have confounded some of the ways in which we as educators have assessed students in the past. 

How can we partner with our students to ensure they are using AI tools ethically and with a critical thinking mindset?

Dr. Anne Arendt-Bunds

I assure you, all is not lost. We can actually use these tools to help enhance both education and assessment. 

This article summarizes just a few of the methods that can be used for student learning and assessment. The intention is to give you ideas and to help you and your students think in new ways. 

The technology is here to stay, so we should learn how to best make use of the tools as they develop.

Use AI as a discussion tool.

Instructors often ask students to tackle open-ended questions, assessing their responses based on thoughtfulness, accuracy, and academic rigor. Nowadays, students can use text generating AI tools like Google Bard, ChatGPT, BingChat, YouChat, or similar to simply copy the AI’s answer as their own.

Programs to catch this type of plagiarism are still catching up, and perhaps always will be. So, instead, why not make generative AI a part of the discussion? 

Encourage students to ask AI questions but then guide them to critically evaluate the results and relate them to their own understanding and experiences.

For example, students could ask: 

I need to write a few paragraphs about the similarities, differences, strengths, and weaknesses of lean manufacturing and Six Sigma process management with an industry or workplace example. What should I say?

After some conversation, have the students rethink the question they asked and formulate new, more intricate questions for AI. Then, again, have them assess the results based on what they are learning in class.

A more detailed prompt might ask:

While both lean manufacturing and Six Sigma process management focus on improving organizational performance by reducing waste and streamlining processes and while both are data-driven to assist in decision making that enhances customer satisfaction, it seems there are also many differences between the two, what are they?

Evaluate the outcomes of various generative AI tools and prompts.

Each generative AI tool is unique, resulting in varied responses to the same prompt. Use this to your advantage. 

Have students try multiple tools with the same question and critically assess the outcomes. Encourage students to rephrase their prompts and repeat the process to observe how the AI-generated responses differ. Discuss the strengths and weaknesses of the information they received. 

Integrate AI with non-AI tasks or multiple AI systems.

Designing a multifaceted assessment that incorporates various resources and tools demands student input, creativity, and problem-solving abilities due to the requirement of using multiple tools.

For example, students may need to combine different content types like text, images, avatars, audio, and video using various AI generative tools such as ResponsiveVoice for text-to-speech or Craiyon for image creation.
To illustrate, students can start by creating a presentation using Powerpoint, SlideShare, Canva, or Prezi. Additionally, they can experiment with Gamma, a tool that formats content into presentations. Afterward, they should review and enhance the slide deck and include a personal avatar with a voice track to provide further insights, culminating in a comprehensive video format.

Have students debate with AI results.

Results from generative AI are not typically static, as they continuously iterate based on machine learning updates. 

Additionally, the results do not always consider factors such as context or subjectivity and may even be giving results based on inaccurate inputs (since really what it is doing is parsing out information from a slew of possible inputs). This is another thing we as educators can use to our advantage. 

By debating with AI, students can learn how to counter argue, improve their critical thinking, and formulate their own viewpoints—what we all want to see in our students!

Use AI to critically reflect on current events or cultural context.

Artificial intelligence technology is not good yet at teasing out cultural context nor is it good yet at very recent events (for which information sources have not yet amply been developed for the AI to tap into). These limitations can be used for student learning by offering a starting point that then necessitates research, self-reflection, and critical thinking.

For example, have students report on a currently open Supreme Court case that has not yet had a hearing and consider it from a specific cultural lens (ethnic group, demographic area, income level, gender identity, sexual orientation, etc.)

Use AI to help generate scenarios.

As instructors, we can use generative AI to help us formulate content, too. 

Consider, for example, if we use it to generate case studies or scenarios based on criteria we specify. We can use these outputs to spark conversation and debate.

I could then take those results and continue to refine the output since the initial output from generative AI is likely to make me a rather generalized case study or scenario. By refining it, I can add localized touches or personalization to make it more relevant to the demographic or scenario.

Use AI to discover and validate alternative solutions.

In many situations, a problem has more than one solution, all of which may be equally valid.

Have students find multiple alternative solutions by using generative AI tools and then select and defend why they chose one of them.

This can even be done with areas of science and math that have multiple approaches, methods, or paths that can be taken to get to the same ultimate answer. 

Consider if you ask a student something like “find the sum of the first 100 positive integers”. To solve it, they could use the formula for the sum of an arithmetic series (S = n(n+1)/2), or they could use a shortcut for the sum of consecutive integers (such as pairing the numbers from both ends and summing the number of pairs). 

The students could then be asked to explain the underlying principles that are key to the method they chose.

Encourage the use of generative AI for idea or design development.

Generative AI can be powerful tools in idea generation, design development, drafting storyboards, and proposing alternatives. This can be used to benefit instructors and students alike. 

Let’s say you want to discuss possible futures for a given topic. This could include things like projected scenario outcomes, business success projections, social or cultural escalation or de-escalations, technology expansion or contractions, etc. In each case, none of us know for sure what is to come. Generative AI could be used for dialogue by asking it what it sees as potential implications, challenges, or opportunities. 

AI may not be perfect at art and design, and perhaps may never be based on the end user perspective, but it is undeniable that it can be used for idea generation at a minimum.

Sometimes we may give an assignment where the student is not sure where to even start.  Again, generative AI could be helpful.  For example, let’s say you ask students to design, build, and document (inclusive of appropriate terminology and symbols) a simple circuit that accomplishes some set of tasks such as ringing a buzzer while also turning on a light. 

Some students may know immediately how they plan to go about it. Others may be frozen in their tracks, not sure where to even start. Generative AI could help students find explanations and solutions which would help lead them to their functional tangible model.

Final thoughts

In our own roles as educators, we should ponder how generative AI can be a useful tool and resource for us as we develop programs, curriculum, content, and assessments. 
Tools like Quillbot, ProWritingAid, LanguageTool, Writefull, or Grammarly, for example, could assist us with our own copy editing or even language conversion. Tools like Tone Transfer or Open AI Jukebox could help us make sound tracks for videos or presentations we may want to make.

For students, we should consider welcoming the use of AI or other tools to learn, but have the assessment outcome be their own. Ask your students to consider generative AI options as tools, just as any other software would be.

In considering the above, and all other aspects of education, we need to contemplate for ourselves what we see to be the primary purpose of teaching and assessment. The responses we have may vary by discipline, course, and even topic.  If we are seeking to have students be capable of finding accurate, relevant, and timely answers quickly, then learning tools or generative AI perhaps should be welcome and even encouraged as starting points.

Whatever the case, consideration of how generative AI or other learning tools may play a role both in education and in industry is likely advisable, informative, and helpful from the lens of not just assessment but curriculum, content development, and lifelong learning.