Recent advances in generative AI technology have made synthetic human personalities a reality. It should be possible to create a proxy for our most likely users and create virtual scenario tests.

Existing Large Language Models are useful for single-thread conversations, however the conversations are very one-dimensional, giving only one perspective at a time. Existing attempts to have LLMs collaborate have been very erratic. If we could find a way for multiple personalities to co-exist in a focus-group style conversation, more value and insight could be derived.