Jon and Catie walk through US Patent 10,452,978, Google's patent on the transformer, the design behind the T in GPT. Starting from what a neural network is, they build up to what claim one covers, what the dependent claims add, and four drafting lessons for anyone patenting artificial intelligence.
Sources:
US 10,452,978 B2, Attention-based sequence transduction neural networks: https://patents.google.com/patent/US10452978B2/en
Claims of US 10,452,978: https://patents.justia.com/patent/10452978
US 12,217,173 B2: https://patents.google.com/patent/US12217173B2/en
Attention Is All You Need (arXiv, version 1): https://arxiv.org/abs/1706.03762v1
Generative pre-trained transformer (Wikipedia): https://en.wikipedia.org/wiki/Generative_pre-trained_transformer
Read the article: https://burbagelaw.com/podcast.html
Narrated with the assistance of AI. Nothing in this podcast is legal advice.
Artificial intelligence patent law, from a registered patent attorney. Every Monday, the week's developments: Federal Circuit decisions, Patent Office guidance, and what the largest AI companies are filing. Every Friday, original data computed from Patent Office records. On Wednesdays, a longer episode on the Supreme Court, the inventions that changed things, and what Shakespeare can teach a lawyer. On Saturdays, one granted artificial intelligence patent, explained from what the technology does to what the patent lets its owner stop others from doing. And a short episode on each new Federal Circuit decision on artificial intelligence patents. From Jon Burbage of Burbage Law, PC, in Salt Lake City. Narrated with the assistance of AI. Nothing in this podcast is legal advice.