Machine learning US 20,260,057,214 A1 Under examination

Incorporating complex product requirements in search ranking system

This invention describes a way to train an AI system to rank content for a network service, like a social media feed or search engine. It uses a powerful first AI model to generate detailed labels for content, following specific natural language instructions that define what 'good' content looks like. These AI-generated labels are then used to train a second, more specialized AI model, teaching it to understand and apply the service's quality objectives.

Why it matters: Since 2024, the capabilities and efficiency of large generative AI models in following complex, structured instructions like Chain of Thought for data labeling have significantly advanced. This makes the process of generating high-quality synthetic training data for specialized models more feasible and cost-effective than it was when filed.

Status: Under examinationA pending application. It has not been granted, and it has not been abandoned.
How hard to build: SpecializedAI model training, data pipeline, NLP expertise

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.

+10