Natural language data generation using automated knowledge distillation techniques

This invention describes a computer system that can answer questions in plain language. When you ask it something, it first searches through a collection of information using artificial intelligence to find relevant pieces. It then ranks these pieces of information, specifically using a type of AI model called a BERT-based model, to figure out which ones are most important, and uses special techniques to create a clear, natural language answer from the best information.

Why it matters: Filed before the widespread adoption of powerful large language models (LLMs), which have since dramatically improved natural language generation capabilities. The techniques described, particularly for generating answers and distilling knowledge, are now significantly more feasible and effective using modern LLM architectures.

Status: Not in forceListed as no longer active. The specific reason is not in the record we hold.
How hard to build: SpecializedComplex AI algorithms, software system

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