The COVID-19 pandemic accelerated the movement and creation of educational material from in-person to online/virtual. In Spring 2021 I was the instructor of record for a section of Calculus II at Vanderbilt University, and the transition from in-person revealed to myself and many other educators the difficulty of creating effective online educational content. Engagement was difficult and the amount of effort needed to convert and deliver our material to be online-friendly easily took at least twice the time it took when in-person. And while in our shock many of us attempted to simply emulate the in-person classroom experience, the reality is that designing and running an online course effectively requires much more.

Though colleges are generally holding classes in-person again, the need for effective online education has not diminished. Online, asynchronous courses are especially relevant to professionals, such as university research staff learning how to use the university's supercomputing resources. Designing effective content for professionals has added difficulties associated with it:

• Larger degree of background knowledge and time since formally taking relevant courses;
• Often have a wide variety of reasons for taking a course and plan to get different things out of that course;
• Have schedules that put constraints on the feasibility of synchronous or in-person courses;
• etc.

For these reasons and others, having consistent and complete data on student behavior can be much more difficult to obtain, making it more difficult to gauge the effectiveness and usefulness of a course.

This is where machine learning and learning analytics come into play. AI/ML tools might be used to discover and classify student needs and background, automatically produce concept maps from existing online content, or recommend appropriate 'next' lessons based on student performance, progress, and determined interests/background.