Facebook is allowing some of the company’s AI Research to be a part of the larger technology collective that exists around the world. The social network has donated certain parts of its AI research to an open source group called Torch. Some of the technology included in the release is focused on improving, and speeding up the rate at which a machine learns. Specifically, Facebook Research points out that one of the new processes that are included in the batch can speed up the machine learning algorithms by 23 times. An astronomical number when you consider how fast that actually is. Companies like Microsoft, or Google though, most-likely won’t be using the technology or research to do anything groundbreaking. Instead, it is quite likely that this research being made public – will allow Facebook to explore more options, by putting more heads inside of the collective pool.
This would effectively allow smaller operations to push and expand their operational limits and find new ways to use that artificial intelligence research and technology, while allowing Facebook to further its own cause by developing that information and using it in a way that is more cost-effective than previous methods. That being said, the cost-effectiveness according to Facebook is something that will be an incredible asset to those who use the research and technologies included. A great example is comparing it in dollars-and-cents.
All operations are executed via code, and what this does is update significantly outdated, or outperformed code – according to Facebook. The company pointed out that for a simple separation of data, or a simple data sort – if it cost a dollar to sort previously, in terms of server time – the new methods would cut that cost down to five cents. A significant upgrade for any company, or any operation that is looking to do things more-efficiently. Which is something many small operations are looking to do.
Facebook though like many other big companies such as Google and Microsoft – don’t necessarily operate to benefit others. At least not typically. However, this offering shows just how much Facebook is willing to do, in an effort to push the overall information that is out there. There’s the idea that this could start a trend, that would see more large companies like Facebook, being a part of the overall networking and computing solution, rather than racing to be the first company to perfect and monetize the results of faster computing.