In 1959 Volvo introduced the modern 3-point seatbelt. Since then it has saved countless lives, there are millions of people walking the earth today who would not be if it had not been from that single invention. What also made it such a revolution was that despite the considerable advantage it would have given Volvo over their competitors, they gave away the design for free.
This sharing is what has allowed the design to save the lives of people driving everything from Fords to Maseratis today. If Volvo had simply kept the design for themselves, they would, in all likelihood, be one of the biggest car companies in the world today. Instead, they are not even in the top 30, but their work has allowed those around them to develop new technologies that have pushed the development of the automobile to a place that few would have envisioned 70 years ago.
Volvo’s choice to share their innovation helped the world, but today this is something that few companies are doing, and larger companies in the AI space are among those to have steered clear.
According to McKinsey's State Of Machine Learning And AI, 2017, ‘Artificial Intelligence (AI) investment has turned into a race for patents and intellectual property (IP) among the world’s leading tech companies.’ Which means that despite these companies pushing the industry forward to some extent, they are essentially trying to hold as much power themselves as they possibly can.
There are examples of companies allowing others to use their work, for instance, Google’s Tensorflow, but the reality is that by concentrating the power in a few hands the speed of development may well be slowed or only developed in specific areas.
When we think back to the success of big data over the past decade, one of the concepts that has been so key in its development has been open source software, promoted by organizations like Apache. However, in many ways, the current situation is the opposite of this and what Volvo did with the seatbelt - companies seem to be grabbing every small patent possible to maximize their advantage over the competition.
This is short-termism at its worst, essentially blocking anybody else from using the smallest part of any software that your company has created, slowing its development in areas outside of your core business. It may well be that Google are using their software for good in many areas, but the reality is that through controlling where this software is being used they are dictating which projects can or cannot use it. When project ‘worth’ is seen through the lens of a particular company it skews the overall use of the technology, limiting its use to those deemed important to whoever owns a particular patent.
The current data ecosystem has been built around collaboration, openness, and sharing, it is one of the key reasons why it has become such a game changer. This isn’t simply about some airy feeling either, it is a relatively firmly established principle. The NSA, which arguably holds the most diverse and sensitive information about US citizens, admitted in 2013 that despite having billions of dollars to spend on data management tools, none they could find could out-innovate Hadoop, purely because it was open-source and thousands of data junkies were constantly making it more powerful.
However, when I say that Hadoop is an open-source platform, it does perhaps miss exactly how the platform was built in the first place.
The idea initially came from a research paper released by Google, which Yahoo! then ran with and created the foundations of what we know as the modern Hadoop today. It is all well and good having thousands of people working on something, but through having private companies utilizing their internal talent to create a truly world changing technology, they pushed it forwards and allowed it to take shape before becoming truly open.
This may well be something that we see with AI technology, with companies spending the time and resources to create something great before releasing it. However, with the number of patents in the area, this would certainly represent a risk. We know from experience that when it comes to patents and competition that companies who would be considered as competitors in the tech space, like Facebook and Google, often do not work well together. For the sake of simplicity, if Facebook hypothetically owned 50% of all AI patents and Google hypothetically owned the other 50%, the chances of them working together to create an open solution is close to 0.
It could potentially create a situation where we have the technical knowledge to create a world changing piece of AI technology, but due to the patents held, the ends won’t meet.
As we move into an AI future, this hoarding of patents also makes moves to standardize practices more difficult. Google could be teaching their systems to read data in one way, whilst Apple could be teaching theirs in another, meaning that joining the two at some point in the future would be considerably more difficult than if they had simply worked together to achieve the same goals.
It is impossible to predict the future, so the fears around the hoarding of patents may be completely unfounded, but for the time being, we need to be sceptical of why this is taking place and the potential implications it may have.