Group Messaging Prompts for Multi-User Generative AI Context
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional generative models are designed for single-user interactions and cannot maintain conversation context in group conversations, preventing their incorporation into computer-implemented messaging applications that support group discussions.
Innovation Solution
Incorporating a generative model into a computer-implemented messaging application by providing a prompt that identifies the conversation as a group conversation, includes participant identities, and utilizes a transformer-based model to generate outputs based on multiple messages, allowing the model to recognize and respond to multiple users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a conventional generative model is used for single-user interaction, then the model can maintain conversation context effectively, but the model cannot disambiguate between different users in group conversations
Solution Approach 1:
The patent segments the conversation context by associating each message with a unique user identifier. The generative model processes messages individually, maintaining separate context tracks for each user within the group conversation. This segmentation allows the model to distinguish between different users while maintaining overall conversation context, resolving the contradiction between group conversation support and context maintenance reliability.
2Adaptability or versatility
If a generative model is incorporated into a messaging application, then new use cases like summarization and question answering become possible, but the model cannot inherently recognize multi-user conversations
Solution Approach 1:
The patent implements preliminary action by pre-constructing prompts that include user identifiers and conversation context before the generative model processes the messages. The system prepares the prompt structure in advance, incorporating placeholders for user identities and message histories, so that when actual messages arrive, they can be directly mapped into the pre-defined prompt framework. This reduces the complexity of real-time prompt construction while enabling expanded use cases.
3Productivity
If the generative model processes multiple messages from multiple users, then group conversation functionality is enabled, but the model cannot disambiguate which user generated which message
Solution Approach 1:
The patent introduces user identifiers as intermediary elements that mediate between the messages and the generative model. Each message is tagged with a user identifier, which acts as a mediator carrying information about the message's origin. The generative model processes these tagged messages, using the user identifiers to maintain accurate attribution and disambiguation. This intermediary mechanism enables efficient group conversation processing while preventing information loss regarding user identities.
Data Source
AI summary
A computing system includes a processor; and memory storing instructions that, when executed by the processor, cause the processor to perform several acts. The acts include receiving multiple messages from multiple users in a messaging application that supports group conversations, where the multiple messages are included in a group conversation. The acts also include providing a prompt to a generative model, where the prompt includes the multiple messages. The acts additionally include receiving, from the generative model, an output generated by the generative model based upon the prompt and including the output as a turn in the group conversation.