Generative Model Search Integration for Factual Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional search engines and generative language models are inadequate in providing accurate and relevant information, especially when reasoning over multiple webpages or requiring up-to-date information.
Innovation Solution
Integrating a search engine with a generative model to provide additional context, where the prompt includes user input, information from viewed webpages, and search results, enabling the generative model to generate more accurate and relevant outputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If a conventional search engine is used to retrieve information, then it can return search results and knowledge cards, but it cannot provide reasoning over content or generate accurate answers requiring computation from multiple sources
Solution Approach 1:
The patent combines a search engine with a generative model to create a hybrid system. The search engine retrieves relevant webpages and content, while the generative model processes this content to generate reasoned answers. This merging allows the system to both retrieve comprehensive information and automatically compute answers, eliminating the need for users to manually process multiple sources.
2Speed
If a generative model is used to generate output based on training data, then it can respond quickly to queries, but it produces factually incorrect information and cannot access up-to-date information
Solution Approach 1:
The system performs preliminary information retrieval using the search engine before the generative model generates its output. By first gathering accurate, up-to-date information from web sources and then using this as context for generation, the system ensures factual accuracy while maintaining quick response times. The search engine pre-processes the information needs, allowing the generative model to work with verified data.
3Adaptability or versatility
If a generative model is trained on static training data, then it can generate outputs based on learned patterns, but it cannot provide meaningful responses to time-sensitive queries like current weather or recent news
Solution Approach 1:
The search engine acts as an intermediary between the generative model and time-sensitive information sources. Instead of retraining the model to access current information, the system uses the search engine to retrieve up-to-date data (weather, news, sports scores) and provides this as context to the generative model. This allows the model to respond to time-sensitive queries without any retraining, maintaining both adaptability and efficiency.
4Measurement precision
If the prompt includes additional context from search results and webpages, then the generative model produces more accurate output, but the prompt size and processing complexity increase
Solution Approach 1:
The system extracts only the most relevant portions of webpage content and search results to include in the prompt, rather than incorporating entire documents. By selectively extracting key information, statements, and data points that directly address the query, the system maintains high output accuracy while keeping the prompt size manageable and processing complexity reasonable.
Data Source
AI summary
A computing system is disclosed that includes a processor and memory. The memory stores instructions that, when executed by the processor, cause the processor to perform several acts. The acts include receiving, by a generative model, input set forth by a user of a client computing device that is in network communication with the computing system. The acts also include generating, by the generative model, a query based upon the input set forth by the user; providing the query to a search engine. The acts further include receiving, by the generative model and from the search engine, content identified by the search engine based upon the query. The acts additionally include generating, by the generative model, an output based upon a prompt, where the prompt includes the content identified by the search engine based upon the query. The acts also include transmitting the output to the client computing device for presentment to the user.