{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"Thinking Makes It So","title":"October 31: The patent behind word2vec","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/fdd1d7ed\"></iframe>","width":"100%","height":180,"duration":854,"description":"Jon and Catie walk through US Patent 9,037,464, Google's patent on the method behind word2vec, which turns each word into a list of numbers. Starting from why a computer needs words as numbers, they build up to what claim one covers, how Google has kept the patent family open since 2013, and four drafting lessons for anyone patenting artificial intelligence.\n\nSources:\nUS 9,037,464 B1, Computing numeric representations of words in a high-dimensional space: https://patents.google.com/patent/US9037464B1/en\nClaims of US 9,037,464: https://patents.justia.com/patent/9037464\nUS 9,740,680 B1: https://patents.google.com/patent/US9740680B1/en\nUS 10,922,488 B1: https://patents.google.com/patent/US10922488B1/en\nUS 2026/0170243 A1 (Application 19/360,947): https://patents.google.com/patent/US20260170243A1/en\nEfficient Estimation of Word Representations in Vector Space (arXiv, version 1): https://arxiv.org/abs/1301.3781v1\nWord2vec (Wikipedia): https://en.wikipedia.org/wiki/Word2vec\n\nRead the article: https://burbagelaw.com/podcast.html\n\nNarrated with the assistance of AI. Nothing in this podcast is legal advice.","thumbnail_url":"https://img.transistorcdn.com/Tj0jgdq-8N-yIwTpxVtvfZh3bDBXinEYJfSSBRBkgKE/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS84Yjcy/NDVkYzg0MTViMjEw/ZmYwMWEzMjAxYmFk/MjcwZS5qcGc.webp","thumbnail_width":300,"thumbnail_height":300}