AI-Driven Academic Operations at Arab Open University — Retention Analytics to Automated Timetabling

Arab Open University (AOU) is a Kuwait-headquartered, pan-Arab distance-learning university enrolling roughly 62,000 students across nine regional branches (including Kuwait, Lebanon, Jordan, Saudi Arabia, Egypt, Bahrain, Oman and Sudan). Two separate AI-driven interventions, documented nine years apart, illustrate how the university has used data and automation to run its multi-country operations: a 2017 rollout of IBM Watson Analytics for retention and quality-assurance decisions, and a 2026 peer-reviewed rule-based AI system that automates academic timetabling.

AOU adopted IBM Watson Analytics, a cloud-based data-discovery and predictive-analytics service, to make sense of student data that had become 'too complicated to understand using only spreadsheets' across its diverse campuses, feeding quality-assurance interventions and a redesign of its introductory mathematics course. Separately, AOU replaced a labour-intensive manual timetabling process with a rule-based, Python-built automated scheduling tool, deliberately choosing a transparent rule-based method over black-box optimisers (e.g. genetic algorithms, simulated annealing) for institutional accountability, and deployed it across all nine branches.

The Watson Analytics-informed interventions reduced average student dropout by 18%; the redesigned introductory mathematics course cut withdrawals by 34% while raising pass rates by 12 percentage points; programme redesign avoided an estimated 6-11% of at-risk tuition revenue; and administrative efficiency improved by an average of 10%. The analysis also revealed unequal facilities and staff-to-student ratios across branches (Lebanon and Oman lagging Kuwait, Bahrain and Egypt), prompting targeted follow-up. The automated timetabling tool cut timetable-building time from four-to-six weeks to under two hours, reduced scheduling conflicts by 85%, and raised classroom utilisation from 65% to 78%.

Both results rest on a single source each — a vendor case study and a single-author peer-reviewed paper — with no independent third-party replication, and neither system's outcomes were broken down by gender, disability or income.