In an open-source effort to democratize ideas, artist, engineer, and internet activist Alexander Reben has created a clever new project called All Prior Art. With this project, Reben’s aim is to deconstruct all published patents and reconstruct an almost infinite number of new inventions from the rubble of the old. He then publishes the work publicly. Among the absurd, empty, and inane ideas will certainly be something of value, which some day someone might want to patent. But since All Prior Art has already published the texts, the new patent application will be invalid.
The task is too tremendous for any human –or even team of humans– to undertake, so Reben created an algorithm to assist him. Tireless and unperturbed by the menial mission, Reben’s algorithm pulls apart texts from the entire database of patents issued in the US and stitches the pieces together again into patchwork paragraphs meant to mimic patent language.
On its first run, The system generated 2.5 million ideas in just three days. The project boasts over 400 volumes — though most ideas make no comprehensive sense. Many read like Dada cut-up poems. Some, however, seem like feasible ideas. According to New Scientist, you can’t patent 3D-printed soap to kill the pests on your strawberries thanks to All Prior Art.
Part of Reben’s goal is to minimize the lawsuits surrounding patent legislation. Patent critics claim the US patent system is a mess. Two-thirds of patent lawsuits in 2015 were brought by “patent trolls,” or people who buy up patents for the sake of leveraging a lawsuit later on. Patent trolls clog up the legal system and make it difficult for “makers” among us to make things because they might infringe on a patent trolls patent. If, however, an idea is proven to have existed –say, in the form of an algorithm’s textual output– before the patent application is submitted, then the patent application will not be accepted.
Reben has asked for support from larger institutions with more resources to run and refine his algorithm, equipping it with advanced methods like machine learning to generate more accurate ideas. Until then his algorithm will mindlessly chug away and publish random texts like a monkey with a typewriter.
- Chess. Jeopardy. Go. Why do we use games as a benchmark for A.I.?
- Exoskeletons with autopilot: A peek at the near future of wearable robotics
- Language supermodel: How GPT-3 is quietly ushering in the A.I. revolution
- The best free Kindle books for 2021
- NFTs and the explosive rebirth of artificial scarcity