2

AI in Education

Case Study: Building Helperbots to Improve Student Learning

Ted Scovell]

Loyola School

Grade 10

Introduction to Programming

SPRING 2025

THE NEW YORK ACADEMY OF SCIENCES

Background

Loyola School

• Located on the Upper East Side of Manhattan
• Serves students in grades 9–12.
• Approximately 50 students per grade
• Approximately one-third of students receive financial aid

​

​

​

​

​

​

​

​

​

[Add photos/graphics/media to help tell your story]

4

THE NEW YORK ACADEMY OF SCIENCES

Problem to Solve

• Students often struggle to debug programs.
• Many work late at night when I’m unavailable to help.
• As a result, some students use LLMs that provide solutions without helping them understand the code—this leads to learning gaps.
• I often answer the same questions repeatedly, so reducing this burden would be helpful.

​

​

​

​

​

​

​

​

[Add photos/graphics/media to help tell your story]

4

THE NEW YORK ACADEMY OF SCIENCES

AI Tools and Applications

I used Playlab.ai to create a chatbot that responds to student questions without giving them direct answers or code. It can:

• Create tutorials on coding concepts

​

• Debug code and provide suggestions

​

• Answer specific coding questions (e.g., ‘How do I load an image in p5.js?’)

​

​

​

​

​

​

​

[Add photos/graphics/media to help tell your story]

4

THE NEW YORK ACADEMY OF SCIENCES

Implementation Strategies

I introduced the chatbot in class and shared a link to it on the course page. To encourage use, I offered extra credit to students who emailed me feedback about the helperBot.

​

​

​

​

​

​

[Add photos/graphics/media to help tell your story]

4

THE NEW YORK ACADEMY OF SCIENCES

Observed Benefits

While there has been enough time to fully assess the impacts of the helperBot, it appears that the helperBot has:

​

• reduced students who turn in late homeworks because they can have their questions answered at night

​

• Helped students solve problems independently, especially at night

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

Image generated using Gemini from the prompt: “make an image of a young woman writing code and smiling”. Notice that the code is appearing on the wrong side of the computer..

4

THE NEW YORK ACADEMY OF SCIENCES

Data from survey: most students are open to trying the helperBot

8

4

THE NEW YORK ACADEMY OF SCIENCES

Data from survey: most students found the helperBot easy to use.

8

4

THE NEW YORK ACADEMY OF SCIENCES

The helperBot helped students complete work on time

8

4

THE NEW YORK ACADEMY OF SCIENCES

Data from survey: most students found the helperBot helpful.

8

4

THE NEW YORK ACADEMY OF SCIENCES

Challenges and Considerations

​

• I remind students that I’m available during school hours, but I’m concerned they may come to rely solely on the chatbot, which could affect our teacher–student relationship.
• Some students may get frustrated with the chatbot because it doesn’t provide code, and may turn to other tools that do.

​

​

​

​

​

​

​

​

[Add photos/graphics/media to help tell your story]

4

THE NEW YORK ACADEMY OF SCIENCES

Future Directions

I currently use the chatbot in my 10th-grade class. Next year, I plan to expand its use to all my classes and offer a tutorial to other teachers interested in creating their own.

​

​

​

​

​

​

​

​

​

[Add photos/graphics/media to help tell your story]

4

THE NEW YORK ACADEMY OF SCIENCES

Key Takeaways for Other Educators

Preliminary data suggest that tailored chatbots could:

• Increase student learning�
• Reduce academic dishonesty
• ​
• Enable more students to complete work on time�
• Allow teachers to focus more on students who need direct support

​

​

​

​

​

​

​

​

[Add photos/graphics/media to help tell your story]

4

THE NEW YORK ACADEMY OF SCIENCES