When a group of individuals attempt to plan a group activity such as a joint trip to a common destination, the presence of conflicting constraints makes it difficult to arrive at a plan that is agreeable to all. This disclosure describes a virtual concierge that accepts as input multiple, potentially conflicting constraints from multiple individuals planning collective travel (or other group activity) and outputs optimized recommendations tailored for the individuals in the group. The virtual concierge application can leverage large language models (LLM) for language understanding and for natural user interactions. The virtual concierge can generate prompts for an LLM that has been efficiently tuned using techniques such as adapter layers, few-shot prompt tuning, etc. Machine learning (ML) can be used to generate a set of recommendations based on the preferences of different individuals in the group.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.