As generative artificial intelligence (AI) continues to grow in popularity, educators face a dilemma: should we use it in the classroom? That’s what School of Information Sciences (SIS) Patricia D. Williams Professor, Dania Bilal, is setting out to learn, at least in part, with a new project funded by a UT System Grand Challenge Grant.
“Most schools still don’t use generative AI because there are many issues, including plagiarism, ethical concerns, bias, and social implications. We understand that, but we can’t wait until these are solved to introduce generative AI to kids,” Bilal said.
Acting as a Principal Investigator (PI) Bilal and a team of three other researchers within the UT System submitted a proposal to work with children in grades seventh to twelfth who are in 4-H. The SIS professor is joined in this project by:
- Professor Mehmet Aydeniz (Co-PI), professor, UT College of Education, Health and Human Sciences
- Associate Professor Jennifer Richards (Co-PI), associate professor and UT Extension teaching and learning specialist, Herbert College of Agriculture, and executive director of Children Learning through Outdoor Experiences
- State Program Leader Jamie Harris, state program leader, Tennessee 4-H and Youth Development, UTIA
Learning how children can use generative AI could prove beneficial to under-served populations, Aydeniz said, who is interested in the intersection of education and equity via AI. Because children will inevitably use generative AI outside of schools, he said it is imperative they learn both the potential benefits and harms such a tool can bring and how it impacts students.
“At the surface level, the tool looks like another tool that will be beneficial to students; but if it’s giving students the answers and they’re skipping the thinking stage, is that really beneficial? If that happens, how do we change our assessments to accurately measure students’ progress in developing critical thinking skills?” Aydeniz queried.
The project will kick off with pilot testing potentially in spring 2025, during which the team will refine their research design before implementing it in the summer. During the summer, they will spend a weekend each at four camps with 4-H members and students, where they will first identify how much the children know about generative AI tools and then observe how they interact with such tools, especially Microsoft Bing Copilot. The children will be given learning-driven tasks related to events or issues in the state of Tennessee to perform utilizing Copilot.
After the team collects the data on a few tasks, they will train the children on how to use generative AI, especially prompting and educating them on how to evaluate the information generated by Copilot and identify bias and misinformation. Then, they’ll ask the children to complete the remaining tasks. The team will assess the children’s learning through observations, interviews, and pre- and post-test questionnaires. From this, they will build an AI literacy framework for educators on how to introduce AI and generative AI tools in the classroom. Bilal said they hope to train 4-H educators at a two-day workshop on generative AI and to implement the framework in their teaching. She noted the goal is for this framework to be adaptable to different learning environments.
“We want to introduce children to the basic AI concepts, and what generative AI is. We want them to be well-educated in that context. What’s important is that they build adequate mental models of generative AI through the training ‘Oh I know how this works.’,” Bilal said. “We want them to know where the information comes from, how generative AI puts it together, and how they can use good “prompts” to get information that they need.”
Richards, who works with 4-H programs throughout the state in her role at UT Extension, said the 4-H members will be excited to participate in research that uses AI—something she knows youth are very interested in exploring.
“When they brought this idea forward, I thought 4-H was a perfect space to look at those types of emerging technologies because it’s a non-formal learning environment and we’re not bound by grades or state testing,” she said, adding the project will also expose the children to career options they would otherwise likely not see or experience.
As AI becomes pervasive and integrated into everyday life, it will become an expectation that the next generation of workers will need to know how to use it. Those who aren’t literate in appropriately using these tools will be left behind, Aydeniz said, which is why teaching it in K-12 grades is ideal.
“The tool, if used correctly, will offer significant benefits to students who are struggling. It serves as a scaffold for them to get over the basic needs of literacy and makes resources accessible to students they wouldn’t otherwise have,” he said.
In addition to conducting the testing, the team has been invited to present their proposal at the UT Grand Challenge Summit in Murfreesboro in February 2025, where they will discuss the design and how it will be implemented.