Well, this morning, Google took O’Reilly’s idea further than even he expected. It’s not selling access to its deep learning engine. It’s open sourcing that engine, freely sharing the underlying code with the world at large. This software is called TensorFlow, and in literally giving the technology away, Google believes it can accelerate the evolution of AI. Through open source, outsiders can help improve on Google’s technology and, yes, return these improvements back to Google.
I told my roommate, who’s taking a lot of economics classes at the moment, about this, and he immediately asked “what’s that going to do to their stock?”
And the fun answer is: not a lot. Because, sure, open-sourcing the tech means that anyone can use it, but very few people actually have the resources to use it at the scale that Google does. Amazon, with their S3 cloud, probably does, but they’re not competing with Google in search. Facebook might, but they don’t do web search, and their internal search stuff is a custom-built solution that works better for what it needs to do than what TensorFlow could produce.1 Microsoft is also in the ‘might’ category, but even if their Azure Cloud could run TensorFlow at the scale necessary to back their search engine, I doubt Bing actually has enough of a budget to justify it.
So, while it seems like a nice gesture, it’s a rather calculated one – Google gets all the benefit of boosting public AI research, all the good PR of putting their code out there for anyone to use, and very little risk that it’ll actually help any of their competitors.
- And it already has all the biases they want it to have, something that would take time and money to train into a new system. ↩