/Patent Tracker
GooglePatent applicationAI & softwareInterest 6/10

Google patents method to create readable fake text for training AI models

Google has filed a patent for a method that creates synthetic, human-readable text specifically designed to train generative neural networks. The system generates this data by optimizing for performance on a specific task while ensuring the text remains easy for humans to read, rather than just being random tokens. This approach allows the AI to learn from high-quality, understandable examples rather than raw, unreadable data.

Why it mattersThis is interesting because it addresses a common challenge in AI development: ensuring that the data used to train models is not only effective for the algorithm but also makes sense to humans, which can improve the overall quality and reliability of the AI's output.

In the filing’s own words

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating human-readable synthetic text for training generate neural networks thereon.

What exactly they claim (claim 1)

1. A method performed by one or more computers and for fine-tuning a generative neural network, the method comprising: receiving a source dataset for a downstream task to be performed by the generative neural network, each sample in the source dataset being a sequence of tokens in a vocabulary of the generative neural network; generating a target dataset for the downstream task by optimizing a performance metric that measures a performance of the generative neural network on the source dataset after being trained on the target data set while constraining each sample in the target dataset to be a respective sequence of tokens in the vocabulary having a readability score that satisfies a threshold; and training the generative neural network on the target dataset for the downstream task.

Facts

Company
Google (Google LLC)
Official title
GENERATING HUMAN-READABLE SYNTHETIC TEXT FOR TRAINING GENERATIVE NEURAL NETWORKS
Publication number
US20260260109A1
Application number
19467696
Filed
Feb 2, 2026
Published
Sep 3, 2026
Inventors
6
Claims
23

Classified under

G06N 3/08 Physics & computingG06N 3/0475 Physics & computing
Read the full filing (official PDF) ↗ Open in the Patent Tracker

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.