Putting the Future On Hold: How Fundamental Patents Stifle Progress

There’s a worrying trend occurring in the world of patents. As Electronic Frontier Foundation’s Daniel Nazer points out, more and more patents are being filed for the application of relatively “fundamental machine-learning techniques” to new areas, such as food consumption. This issue represents a crossroads for the U.S. Patent and Trademark Office (USPTO). While the issue of simultaneous invention has been around since Isaac Newton and Gottfried Leibniz independently developed the field of calculus, never before in history has technology and learning advanced as rapidly as it is today. In the past, groundbreaking discoveries took years, even decades, to foment; today, groundbreaking discoveries—particularly in the technology sector—seem to occur every single day. What this means is that the chances of multiple people arriving at the same discovery is significantly increased. The more discoveries there are, the more likely it is that two people find the same one. This is especially true, as Nazer worries, for some foundational techniques in machine learning and artificial intelligence, likely two of the most important technology fields in the next few decades. The USPTO has a significant responsibility to be selective in their acceptance of patents for these emerging technologies. The issue is the same as it has always been: granting a patent may stifle not only competition, but technological evolution too. If a patent is granted for every miniscule new application of the basic techniques of machine-learning, the first to any new application will own the field. There is, of course, value in this—if it’s earned. If an inventor truly discovers a new way of understanding or applying machine learning, a patent is...