“I am prioritizing AI fluency in class so my students will have the proper framework to interact with the emerging AI tools that are currently reshaping modern industry workflows.”

Leslie Hendrix is a Clinical Professor in the Management Science department at the Darla Moore School of Business at the University of South Carolina. Her primary role is coordinating and teaching the second required statistics course for business students. Known for piloting new educational technology, she was one of the first to tackle the challenge of integrating R coding into the business curriculum and later led the charge in developing policies to address the mainstream adoption of Generative AI tools in her classroom.

The Clash Between Easy Solutions and Protecting Core Coding Skills

Leslie’s primary challenge was integrating essential R coding knowledge into a high-enrollment business statistics course while addressing the emergence of Generative AI tools like ChatGPT.

Initially, the instructors used Excel, but Leslie realized its limitations when trying to cover advanced statistical concepts like logistic regression: "I knew I couldn't teach logistic regression full treatment if I was using Excel." Therefore, she decided to adopt teaching basic R coding even though most business students had no prior coding exposure and found even a basic program "pretty terrifying".

When AI became mainstream, students immediately started using Gen AI tools for homework. This led to problems where students would submit code in a different format than what was taught. Leslie mentioned: "Students didn't understand why their outputs were wrong and even didn’t know to check that their outputs were wrong". 

Rather than trying to stop the unstoppable, Leslie decided to move past a policy of banning AI and instead focused on how to use it as a powerful tool for learning. The primary challenge for Leslie was finding a way to integrate responsible AI tools usage by students in her coding classes. She needed a strategy that leveraged AI to teach modern coding skills without compromising her students ability to grasp course material or verify the accuracy of their own work.

Leaning Into AI to Build Better Coders

Leslie decided to fully embrace leaning into AI by teaching students how to use it correctly, with CodeGrade at the center of their strategy. CodeGrade enabled her to address both the structural coding challenges and the responsible AI usage. 

First, the assignment design ensures students use enough of the core course content to be able to “diagnose errors and check outputs”. Leslie structures her assignments in a way that students have to submit their code and take the feedback if the code has failed to improve. This way students have to understand and reflect on the error.

Furthermore, Leslie states that: “By requiring students to use the AI tool built into CodeGrade, the instructors can see the chats.” This visibility allows them to analyze the student’s thinking process and offer specific feedback on how to improve their interaction with the AI. The instructor focuses on two primary skills: prompting and checking results. 

Leslie chose to not put any guardrails on CodeGrade’s AI assistant. This decision was deliberate, as Leslie's goal is to “teach students how to use the full on product properly because that's what they're going to have out in the real world".

Preparing for the Future: Bridging AI and Coding

Leslie believes that the whole university and in particular, the Darla Moore School of business, is actively preparing students for the modern workforce by integrating responsible AI usage in their curriculum. From her perspective the practical knowledge of using AI tools is increasingly common in job postings these days.

CodeGrade enables highly targeted feedback. Instructors can check the auto-test results, the student's submitted code, and their chats with the AI tool to pinpoint the root cause of an error. For example, Leslie mentions that an instructor might advise a student that if they "had asked this question with just these two other words in it, you might have gotten the answer that you needed". This directly teaches the crucial skill of prompt engineering.

The student reception has been very positive. Students who were already using AI tools gave "really good feedback" about the new policy that allows and teaches correct AI usage. The instructor received around 10 unsolicited positive messages from students in a single week which is an incredible achievement for a required statistics course!

Finally, the results show high student engagement. Leslie estimates that from now on her students are trying to "hack a solution" (meaning figuring it out) instead of "game a solution". This strongly indicates that the majority of students are making "an honest effort to try to learn the material".