What important factors should developers consider when designing ChatGPT interactions in their e-learning courses?
First of all... do your users need chatGPT interactions ?
Then remember that AI has its limitations. If you use chatGPT as is, it will search all available resources and the results might be not what your user wants. chatGPT can hallucinate, thus giving inproper results. All results should in the end be doublechecked, only the most simple calls are rocksolid.
Suppose you want to limit chatGPT to use only specific sources or make an asssistant ?
All that is possible with the openAI API. But it is no simple task. You do need to make documents/resources available to chatGPT/openAI to use as source for your calls.
I am working on a chatGPT Assistant at the moment. I will be making a Assistant Builder alongside with it, so our teachers can create their own assistants on our LMS ( CanvasLMS ).
chatGPT as you use it on chatGPT's site is a webinterface build for non-developers to create prompts and use as is. You can build Assistants in that part. Using the API of openAI is the way to add Assistants or chatGPT integration in your elearning projects.
Here you can find some of the steps in this process.
https://www.linkedin.com/posts/mathnotermans_tls-tilburguniversity-chatgpt-activity-7171152604616355841-FgZ8?utm_source=share&utm_medium=member_desktop
The next step is making sure chatGPT remembers the instructions and questions. Will post that soon.
