Query evaluation in natural language processing systems

This invention describes a computer system that answers a natural language question by first sifting through many documents. It uses a large language model (like ChatGPT) to identify the most relevant parts of those documents. Then, it takes only those relevant parts and asks the large language model again to generate a final answer to the original question, potentially by breaking it down into smaller, intermediate questions first.

Why it matters: Filed when LLM capabilities for multi-stage reasoning and precise document chunk evaluation were still rapidly evolving. Today, LLMs are far more adept at these specific tasks, making the system more feasible and effective.

Status: Under examinationA pending application. It has not been granted, and it has not been abandoned.
How hard to build: SpecializedSoftware logic, LLM integration

3 AI reinvention directions for this patent

See all three

AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.

Reinvent this with AI

Where this one came from

Every morning we go through millions of abandoned patents and pick out 15 worth your time. This was one of today’s.

This one

+10