Most people remember their childhood tutors fondly, maybe as positive and supportive figures in their lives. However, as all humans, all tutors have limits. They can’t answer your questions outside of their working hours - Try calling them at night and you’d probably get an earful. 

But not with AI.

Imagine an AI-powered tutor that is available 24/7, clears your doubts, and prepares you for real-life interviews. If you’re a student at Walden University in the US, you can access one such tutor: Julian!

‍

The challenge

Walden University recognized the need for a more personalized and scalable tutoring solution to support its growing student population. Traditional tutoring methods, while effective, are often limited in their ability to provide individualized attention to each student, especially in an online setting. Walden sought to create an AI-powered tutor that could adapt to each student's unique learning style, knowledge level, and pace, ensuring that every individual receives the support they need to succeed.

‍

How was Julian built?

Julian, the most advanced of the three AI tutors at Walden University, was developed with Google Cloud’s AI and ML capabilities. The goal was to create a tutor that moved away from pre-programmed, scripted interactions to something truly dynamic and tailored to the student, offering a range of capabilities aimed at enhancing student learning with on-demand support.

One of the main challenges was developing an AI that could operate without pre-programmed scripts. The team wanted Julian to ingest content on its own, make sense of it, pull key concepts, and foster real tutoring sessions with students. Handling subjective content, such as nursing and education, posed another significant challenge. Unlike many existing AI tutors that focus on subjects with clear right or wrong answers, Julian needed to navigate and provide guidance on more subjective, nuanced content.

Creating a system that could scale easily across different subjects and potentially beyond higher education required developing a modular platform. This platform could automatically ingest, label, and index content to create better dialogues. Ensuring the accuracy of AI-generated content and indexing involved a rigorous process where instructors could review and verify the learning units defined by the AI. This is why balancing AI capabilities with human expertise was crucial to ensure Julian’s effectiveness.

‍

Capabilities of Julian

• Dynamic and Personalized Tutoring - Julian adapts to each student's learning pace, style, and needs. It digests entire courses and curriculum, extracts key concepts, and generates its own questions to foster real tutoring sessions without pre-scripted interactions.
• 24/7 Availability - It provides round-the-clock access, fitting into students' busy schedules, regardless of location.
• Engagement Through Dialogue - Engages students in dialogue via chat functionality, offering different interactions each time to reinforce learning and mastering concepts through practice.
• Learning Activities - Julian generates multiple-choice questions on specific sub-areas. It further evaluates student-highlighted content and creates educational notes for references, all while generating statements for students to paraphrase based on feedback. This offers a method to test the student’s knowledge with personalized questions. 
• Feedback and Evaluation - Provides immediate feedback, helping students understand and correct mistakes in real-time.
• Subject Versatility - Handles a wide range of subjects, from early childhood education to complex topics like sociology and counseling, making it a versatile tool in the educational toolkit.

‍

Pilot study and results

Walden University conducted a pilot study with over 60 students to test the effectiveness of Julian. The results were overwhelmingly positive:

• Students appreciated the flexible and easy access to the personalized tutoring provided by Julian.
• The tutor offered a different interaction every time, keeping students engaged.
• Surveyed faculty also expressed appreciation for how Julian automated some practice work, allowing them to focus more on teaching and mentoring students. 

The pilot helped Walden validate the effectiveness of Julian before scaling it up more broadly across courses and subjects. They saw Julian as a way to provide personalized learning and tutoring at scale to their diverse student population of working professionals.
‍

As Walden's CIO Karthik Venkatesh noted,
‍

However, every new technology has its own set of limitations. 

‍

Challenges faced with Julian

• Technical Complexity - Developing advanced natural language processing (NLP) algorithms to accurately understand and respond to student queries required extensive training on diverse educational materials. Ensuring real-time interaction demanded robust backend infrastructure to support seamless communication between students and the AI. Integrating these systems with the university’s existing digital platforms, such as Learning Management Systems (LMS), further complicated the technical architecture.
• Data Privacy and Security - Ensuring data privacy and security was another significant hurdle. As an AI tutor, Julian has access to sensitive student data, including personal information and academic records. Protecting this data from breaches and unauthorized access is paramount. Compliance with regulations such as FERPA (Family Educational Rights and Privacy Act) in the United States imposed strict requirements on handling and sharing information. Establishing trust with students and faculty by demonstrating a commitment to privacy and security was crucial for the widespread adoption of Julian. 
• Pedagogical Alignment - Aligning Julian’s capabilities with effective educational strategies also presented challenges. The AI must not only provide accurate information but also engage students in a manner that promotes learning and comprehension. This involves understanding different learning styles and adapting its teaching methods accordingly. Collaborating with educators to refine the AI’s teaching strategies and content continuously is essential to maintain its relevance and effectiveness.
• AI Hallucinations - AI hallucinations, where the AI generates incorrect or nonsensical responses, are a common issue in AI applications. To mitigate this, Walden University employs proper curation and human supervision. The content fed to Julian is chosen by Walden’s researchers to ensure accuracy. Faculty members regularly review the AI's performance and make adjustments as needed to correct erroneous responses. This system of checks and balances helps maintain the reliability of the AI tutor, but severely increases resource requirements.

‍

Key takeaways

The successful integration of Julian at Walden University demonstrates significant potential for AI in education.

• Dynamic and Personalized Learning - AI tutors like Julian provide personalized, round-the-clock support tailored to individual student needs. As time progresses, this is going to become even more relevant with a demand on education services for different learning styles of each student. 
• Balancing Technology and Human Expertise - The importance of maintaining a balance between leveraging AI capabilities and ensuring human oversight. While Julian offers immense support, human tutors and educators play a crucial role in addressing complex queries and providing emotional and psychological support.
• Addressing Challenges - Overcoming technical complexity, data privacy concerns, and pedagogical alignment are critical for successful AI implementation. Aligning AI-generated content with pedagogical standards ensures that the learning material is both relevant and effective.
• Continuous Improvement - Ongoing refinement and adaptation based on student feedback and technological advancements are essential for the AI's effectiveness. This ensures that the AI tools can evolve to meet the changing educational needs and standards.
  

‍

Julian represents a significant advancement in personalized learning and support, promising further advancements in the educational field. As AI technology continues to evolve, the potential for tools like Julian to transform education becomes increasingly apparent. 

‍

Join the conversation! Share your thoughts on AI in education on our social media channels or via email.

‍