This project will eventually have to determine the priority of the suggestion to return to the user based on the transcript, it also utilizes domain-specific languages (DSL) with ai throughout. The goal of this R&D is to determine if utilizing quorum (multiple AI vote), or just a single layer of review could help to significantly reduce issues in DSL format and suggestion priority accuracy.
So for example, we might use a domain specific language to communicate to the frontend via AI how it should be laid out with each response. A hallucination could result in a failure to correctly implement the domain-specific language. Determine how frequently that might happen, then determine if a quorem or review layer would affect the overall accuracy.