What are the most effective ways to implement PBL in online learning?

PBL gets only as good as the level of interactivity you provide your users. Even before defining the problem to solve, it is adamant to quickly get real on the technical capabilities of your tech stack: - what is the level of interactivity supported and intended? - does it support multiple branching scenarios? - do you intend to have a real-time multiplayer approach? Based on these answers, you won't select the same problem.

In my experience, the best learning takes place when students take responsibility for most of these steps by themselves, while being able to come back to the instructor whenever running into a problem. Staying in a close feedback loop and defining assessment criteria + milestones as an instructor while students organize themselves with their resources and constraints is much closer to a real world project than it would be otherwise. This also gives students the chance to tailor each project to their own interests and needs and individualize the flow of work to their own preferences.

It all depends on the level of the learners and the pedagogical objectives pursued. Sometimes it's best to break down the project into clearly identified subtasks, but in some cases finding a good breakdown of the project is part of the statement. In real life, future graduates will have to be able to identify by themselves the different sub-tasks to be carried out.

A great way to strengthen social connections during a teaching session is to ask learners to complete assignments in small groups and then present them to their peers, especially in the form of a short oral presentation and a small written report that will be taken into account for the final assessment. Ideally, you should schedule several discussion sessions with each group, face-to-face or remotely, at the end of the main stages of the work. They allow the teacher to discuss with each of the group members, answer their questions, help them organize themselves and show interest in the work done by the group, which is an important motivating factor.

With access to generative AI to all, the assessment of students working according to the problem based learning approach requires special precautions, because they only need to ask their question to an AI like ChatGPT to obtain a very elaborate answer, which they then only have to deepen, possibly continuing to use AI. Under these conditions, it is no longer necessary for them to carry out the work of self-reflection that is at the heart of this approach. It is therefore preferable to assess them orally by checking in with them several times, face-to-face or remotely, in order to monitor their progress and verify their personal involvement, and to keep a record of it, which can be used in the event that the student disputes the grade.

The risk with some asynchronous pedagogies such as problem-based learning or enquiry-based learning is that learners have gaps in their knowledge of the subject taught. They have generally delved deeper into the subjects assigned to them than those who follow other pedagogical approaches, but they may overlook certain points that were not necessary for the resolution of their subjects. That's why I think it's best to use these two approaches as a way to motivate learners by asking them to work on concrete topics first, and then have them study the full course in a second step, using other pedagogies like flipped classroom. This ensures that they have studied both the entire course and one or more applications.