Abstract
Response suggestions to incoming messages is a popular feature of email and chat/messaging applications. However, while such applications support insertion of images, current response suggestion mechanisms are limited to text. Inserting a suitable image in a conversation currently requires the user to search for a suitable image and copy or download it. This disclosure describes the use of a large language model (LLM) to suggest contextually appropriate images that users can insert quickly within their conversational responses. With user permission, recent conversation history is analyzed by the LLM to identify and suggest contextually appropriate images relevant to the conversation based on key token words from the most recent messages. Further, AI-based image generation techniques can be utilized to generate synthetic images as suggestions. The contextual relevance of the image suggestions can enrich conversations in email and/or chat/messaging applications. The LLM and generative AI models can be implemented on device or on a server and can be part of the messaging application or can be provided as a plugin or separate application.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Biran, Daniel, "LLM and Generative AI Suggested Contextually Relevant Images in a Conversation", Technical Disclosure Commons, (September 15, 2023)
https://www.tdcommons.org/dpubs_series/6253