Artificial Intelligence Patent Landscape 1976-2023

13,244,037 granted patents and PGPubs published from 1976 through 2023 are classified into eight AI component technologies using machine learning models. The dataset provides a detailed landscape of AI innovation over nearly five decades. It was created by the Department of Commerce for academic and research use.

Use Cases

• Analyze the distribution of patents across eight AI component technology categories over time.
• Track the evolution of AI innovation from 1976 to 2023 using patent grant and PGPub publication dates.
• Identify trends in specific AI technology areas, such as Machine Learning or Artificial Intelligence, within the patent corpus.
• Compare patenting activity between granted patents (Patent Grant) and pre-grant publications (PGpub).

Strengths

• Contains 13,244,037 patent records, offering a substantial scale for analysis.
• Covers a long temporal range from 1976 through 2023, enabling longitudinal studies.
• Patents are classified using state-of-the-art machine learning models into eight distinct AI technology categories.
• Sourced from the authoritative Department of Commerce, ensuring data reliability.

Limitations

• The specific features and columns used for the AI classification are not documented, limiting reproducibility.
• The dataset's classification model may have inherent biases or errors that are not quantified.
• The geographic scope of the patents is not specified, which may limit analyses focused on regional innovation.

Provenance

Source

Department of Commerce

Collection Method

Patents classified into eight AI component technologies using state-of-the-art machine learning based models.

Time Range

1976 through 2023

Freshness

Data is current through 2023, with a metadata update in September 2025.