/Patent Tracker
GooglePatent applicationAI & softwareInterest 7/10

Google patents AI answers that automatically include source quotes

This filing describes a method where a language model generates an answer to a question along with a specific quote from a source document that supports that answer. The system uses identifiers to link the generated response directly to the relevant text in the original context.

Why it mattersThis approach aims to make AI-generated text more trustworthy by automatically providing visible evidence for its claims.

In the filing’s own words

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating output sequences using language model neural networks. In particular, the output sequences include a response to an input query and inline evidence that includes a quote from a context document that supports the response.

What exactly they claim (claim 1)

1. A method performed by one or more computers, the method comprising: receiving an input text query; obtaining one or more first context text sequences and a respective natural language identifier for each of the first context text sequences; generating a first input sequence that includes the input text query, the one or more first context text sequences, and the respective natural language identifiers for each of the one or more first context text sequences; processing the first input text sequence using an auto-regressive language model neural network to generate a first output text sequence that comprises: (i) a first output text sub-sequence that is a response to the input text query; (ii) a second output text sub-sequence that is one of the respective natural language identifiers for the first context text sequences, and (iii) a third output text sub-sequence that is text from the first context text sequence identified by the natural language identifier in the second output text sub-sequence; and providing at least the first output text sub-sequence and the third output text sub-sequence in response to the input text query.

Facts

Company
Google (DeepMind Technologies Limited)
Official title
GENERATING OUTPUT SEQUENCES WITH INLINE EVIDENCE USING LANGUAGE MODEL NEURAL NETWORKS
Publication number
US20260260120A1
Application number
18846910
Filed
Mar 16, 2023
Published
Sep 3, 2026
Inventors
5
Claims
20

Classified under

G06N 3/092 Physics & computingG06F 16/33295 Physics & computingG06F 40/284 Physics & computingG06N 3/09 Physics & computing

More from Google in AI & software

Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.