Case Study 5
The Right Course, Right Away: AI Fast-tracks Student Learning Paths for a Prestigious British Public Research University

Personalizing course recommendations at scale with generative AI

The Question: Was there a way to automate and personalize course recommendations without the manual effort?

Requests received via email lacked specificity and consistency.
The team manually searched multiple platforms to find matching courses.
High response times negatively impacted user experience and efficiency.
No unified system for semantic matching of course requirements.

About the Client

The client, a prestigious British public research university offering distance and online courses for undergraduate and postgraduate students, needed a scalable and efficient way to map student and institutional learning requests with relevant courses. Their manual systems were time-consuming and struggled to meet the demand for personalized recommendations. Their challenges included:

The Answer

AI-based Recommendation Engine: Leveraged GPT-3 to suggest semantically relevant courses.
Comprehensive Course Catalog Integration: Over 1,000 courses aggregated from multiple repositories.
Bilingual Testing: The system supported both English and Welsh languages.
Future Enhancement Capability: Built-in features to support additional languages using GPT-3's capabilities.
Automated Mapping: Replaced the manual, platform-hopping search process.
Seamless Workflow Execution: Streamlined process from request intake to course delivery suggestions.

To address these challenges, Infosys partnered with the university to create a generative AI-powered course recommendation engine. This scalable AI solution was designed to streamline and enhance the matching process.

The Outcomes

The implementation of AI has driven significant improvements in efficiency and accuracy for the university's course recommendation process.

85% reduction in turnaround time, resulting in faster and more accurate responses.
Greater relevance due to the shift from keyword-based matching to semantic recommendation for improved personalization.
90% model accuracy, resulting in increased confidence in recommendations provided.
30% potential FTE savings due to reduced manual workload, enabling staff to focus on strategic tasks.

The institution now delivers tailored course recommendations quickly and with high accuracy by combining automation with cutting-edge semantic understanding, setting a benchmark for efficiency in higher education.