As AI advances and technologies like generative AI are transforming the education sector through platforms like Khanmigo, we have observed a significant uptick in demand from learning organizations. Universities are curious whether and how they can build their own AI learning amigo to improve student learning experience and outcomes.
In reality, building a custom version of Khanmigo is more than meets the eye, even with the most powerful generative AI technologies like GPT-4. Here are three must-haves for learning organizations considering building a custom AI chatbot tutor to improve the student learning experience.
1. An AI chatbot tutor should speak the language of individual instructors
Every institution has unique, proprietary content, such as learning curriculums, course materials and assignments. Moreover, each instructor teaches differently. To deliver the best learning experience for students, universities must allow their instructors and staff to build an AI chatbot tutor customized to its organization-specific learning content and follows the individual instructor’s teaching practices.
This customization is vital to reduce or prevent AI hallucinations, which can present false information to students, while still delivering a unique learning experience to the student. Whether a learning organization builds a custom AI tutor chatbot from scratch or partners with an AI company, organization leaders should ensure that their instructors can customize an AI tutor to speak in their unique learning language and follow their teaching flow.
2. An AI chatbot tutor should collaborate with human learning experts
AI chatbot tutors are never meant to replace human teachers or instructors. Instead, they augment educators’ time and allow them to perform more meaningful tasks (e.g., design new teaching methods or advise students). Moreover, AI is far from perfect, and it will make mistakes from time to time. It is thus a best practice to have an AI chatbot tutor collaborate with human learning experts side-by-side rather than work alone.
In the learning context, human-AI collaboration entails two main collaborative tasks:
First, human learning experts should be able directly—instead of going through an IT team—to teach an AI tutor chatbot the initial knowledge, such as course information and lecture notes, as well as to customize the teaching flow as needed. In this process, the AI should be able to provide human experts with instantaneous feedback as to how much it has learned. This type of human-AI collaboration enables the rapid customization and setup of a custom AI tutor chatbot.
Second, an AI tutor chatbot should always keep human experts informed about its performance, like what kind of questions students have asked and how it has responded or not responded. This will enable human experts to improve the AI tutor chatbot continuously. Ideally, such improvements should be made without interrupting ongoing student-AI engagements while allowing the AI tutor chatbot to gain new capabilities and improve its performance rapidly.
3. An AI chatbot tutor should be a unique learning amigo for every student
Instead of being just a passive Q&A machine that provides knowledge and gives step-by-step tutoring instructions, an AI chatbot tutor should be made to serve as each student’s learning amigo to truly impact a student’s learning journey and success. Such an AI learning amigo should automatically infer a student’s unique characteristics, such as their cognitive style, learning interests and preferences and temperament through its natural interaction with the student.
By getting to know a student’s psychographic characteristics, which have been shown to affect student learning success, an AI chatbot tutor can be genuinely empathetic and proactively personalize the learning experiences for each student at scale.
For example, based on a student’s cognitive style, an AI chatbot tutor can deliver custom learning instructions that match the student’s learning style and proactively ask the student questions to encourage deeper learning and critical thinking. Likewise, an AI chatbot can generate homework assignments or quizzes based on a student’s passions, which will better interest and connect with the student and potentially reduce cheating because of the uniqueness of each assessment.
| Chatbot: Hey, fellow baseball enthusiast! Get ready to blend your passion for the game with some exciting math. Imagine a baseball stadium is building a player exhibition wall that echoes the energy of a baseball stadium. The wall takes the shape of a colossal baseball card, 20-foot long, 8-foot high, and 6-inch thick. Now, here’s the pitch: the total cost for this project was $3,840. Can you step up to the plate and calculate the cost per square foot? | Chatbot: Hello, fellow art lover! Get ready to blend your creativity with some exciting math. Imagine an art gallery crafting a wall to introduce a young artist’s work. The wall takes the shape of a giant canvas, ready to be adorned with your favorite brushstrokes. Now, let’s switch to the numbers: the total budget for covering this 20-foot long, 8-foot high, and 6-inch-thick masterpiece with fabric was $3,840. Can you step into the studio and calculate the cost per square foot? |
| Inquiry: An AI-generated, human-edited math quiz for a student who loves baseball. | Inquiry: AI-generated, human-edited, the same math quiz for a student who loves art. |
Best practice must-haves
As an AI tutor chatbot engages with students, it also offers human educators continuous, deep insights into their students. Such insights enable human educators to get to know their students as unique individuals at scale, tracking student progress, identifying at-risk students and taking the next best actions to help students in need.
The rise of AI learning tools like Khanmigo has propelled many educational institutions, including universities, to consider using AI to improve student learning experiences and success. To create a compelling and empathetic AI learning amigo to boost student success, keep in mind the three best practices or must-haves:
- The ability to customize an AI chatbot tutor with organization-specific learning content and instructor-specific teaching practices;
- The ability to support the close collaboration between an AI chatbot tutor and human learning experts;
- The ability to make an AI chatbot tutor every student’s learning amigo by automatically, continuously inferring student psychographic insights and using such insights to help students in a personalized and timely manner at scale.
