Making AI understandable and accessible for students
Liran Ma is building hands-on learning modules to expand students’ comfort level with using AI
•
Published
Liran Ma, chair and professor of Computer Science and Software Engineering, is developing a series of hands-on learning modules aimed at removing barriers and making AI education accessible and understandable for all.
Making AI understandable and accessible for students
•
Published
Artificial intelligence (AI) is rapidly transforming classrooms and workplaces alike. Liran Ma, chair and professor of Computer Science and Software Engineering, is developing a series of hands-on learning modules aimed at removing barriers and making AI education accessible and understandable for all.
The U.S. National Science Foundation has awarded $400,000 to three institutions, with Miami University as the lead institution receiving $200,000, to expand access to AI education through experiential learning opportunities.
“AI is rapidly evolving and has the power to transform education, learning, and the future of work, reshaping the way we teach, acquire knowledge, and engage with technology,” Ma said.
A key to preparing students for an AI-driven world is ensuring that everyone has the opportunity to engage with and learn about AI, according to Ma. He pointed out that traditional AI courses often emphasize complex mathematical foundations, creating barriers that limit accessibility for many students.
With this grant, Ma aims to increase accessibility to AI by enabling more students to use and understand AI tools through experiential learning rather than mathematics alone. The goal, Ma said, is to “lower barriers,” helping students become comfortable with machine learning and bringing new perspectives into the AI discourse.
“We aim to create highly engaging, hands-on meaningful activities with minimal mathematical barriers, allowing students to fully immerse themselves in AI learning across various subjects,” Ma said.
To broaden accessibility to this baseline understanding of machine learning, Ma is designing lab modules that can be used by a wide variety of professors and courses. Ma said he hopes his work can eventually extend into liberal arts courses as well.
One of the lab modules focused on exploring better ways to utilize large language models (LLMs) such as ChatGPT. You can interact with a LLM directly with natural languages, but that can lead to inaccurate results or hallucinations. Hallucinations in AI describe outputs that are factually incorrect or are misleading, including instances of the model making up information, outputting illogical sentences, or giving answers that are impossible.
“To address this issue, I want students to learn techniques, such as ‘magical spells’ (advanced prompt engineering) and retrieval-augmented generation (RAG)," Ma said.
His lab module is designed to create an understanding of “why hallucinations are a feature” in LLMs. Students can become better users of the technology by learning how to work with the LLMs to mitigate or eliminate inconsistencies and incorrect outputs.
“I want people to go beyond just chatting with AI tools like ChatGPT and instead learn how to interact with them more effectively — leveraging their capabilities efficiently, securely, and strategically to enhance productivity, creativity, and problem-solving.” Ma said.
Instruction and implementation will take place here at Miami University, Georgia State University, and Texas Christian University. However, as the project’s main focus is increasing accessibility to AI, the learning modules will ultimately be available online for public download and testing, hopefully within a year, according to Ma.
The U.S. National Science Foundation has awarded $400,000 to three institutions, with Miami University as the lead institution receiving $200,000, to expand access to AI education through experiential learning opportunities.
“AI is rapidly evolving and has the power to transform education, learning, and the future of work, reshaping the way we teach, acquire knowledge, and engage with technology,” Ma said.
A key to preparing students for an AI-driven world is ensuring that everyone has the opportunity to engage with and learn about AI, according to Ma. He pointed out that traditional AI courses often emphasize complex mathematical foundations, creating barriers that limit accessibility for many students.
With this grant, Ma aims to increase accessibility to AI by enabling more students to use and understand AI tools through experiential learning rather than mathematics alone. The goal, Ma said, is to “lower barriers,” helping students become comfortable with machine learning and bringing new perspectives into the AI discourse.
“We aim to create highly engaging, hands-on meaningful activities with minimal mathematical barriers, allowing students to fully immerse themselves in AI learning across various subjects,” Ma said.
To broaden accessibility to this baseline understanding of machine learning, Ma is designing lab modules that can be used by a wide variety of professors and courses. Ma said he hopes his work can eventually extend into liberal arts courses as well.
One of the lab modules focused on exploring better ways to utilize large language models (LLMs) such as ChatGPT. You can interact with a LLM directly with natural languages, but that can lead to inaccurate results or hallucinations. Hallucinations in AI describe outputs that are factually incorrect or are misleading, including instances of the model making up information, outputting illogical sentences, or giving answers that are impossible.
“To address this issue, I want students to learn techniques, such as ‘magical spells’ (advanced prompt engineering) and retrieval-augmented generation (RAG)," Ma said.
His lab module is designed to create an understanding of “why hallucinations are a feature” in LLMs. Students can become better users of the technology by learning how to work with the LLMs to mitigate or eliminate inconsistencies and incorrect outputs.
“I want people to go beyond just chatting with AI tools like ChatGPT and instead learn how to interact with them more effectively — leveraging their capabilities efficiently, securely, and strategically to enhance productivity, creativity, and problem-solving.” Ma said.
Instruction and implementation will take place here at Miami University, Georgia State University, and Texas Christian University. However, as the project’s main focus is increasing accessibility to AI, the learning modules will ultimately be available online for public download and testing, hopefully within a year, according to Ma.
Established in 1809, Miami University is located in Oxford, Ohio, with regional campuses in Hamilton and Middletown, a learning center in West Chester, and a European study center in Luxembourg. Interested in learning more about the Department of Computer Science and Software Engineering? Visit the website for more information.
