Artificial Intelligence is the buzz word du jour, but the technology whose roots go back to research in the 1950s has finally come of age — and it owes a debt of gratitude to cloud computing.
One of the main things that's held AI back up until recently was that it required an inordinate amount of compute power — and that used to be expensive. Certainly in the early decades from the 1950s-1970s compute power was a scarce resource, and getting time on a mainframe was costly and difficult.
Today, those kinds of restrictions are gone. With public cloud infrastructure services like AWS, Microsoft, Google, and IBM; oodles of compute power is available at your fingertips. All you need to do is share your credit card and you're on your way, and you only pay for what you use. This helped solve a major roadblock.
But being able to crunch the numbers was only part of the problem. Companies wanting to build AI into their applications had another big hurdle. It took really smart people with a specific set of skills, and there are a limited supply of data scientists and programmers out there who can delve into AI. If you aren't a big company like Google or Facebook, you could be left out.
Other key ingredients
That means it requires tools to do a lot of the heavy lifting for you. APIs like Watson Natural Langague Processing, Amazon Polly, and Google Cloud Video Intelligence, put these advanced technologies in reach of just about any programmer. As they build an application, developers can tap into these algorithms and many like them, as appropriate. It's incredibly powerful.
Once you have solved the compute and algorithm problem, the next issue is the one of data. That's a tougher nut to crack, but the cloud can help here too with sites like Kaggle or the UCal Irvine Machine Learning Repository. Finding the right data could be a bigger challenge though.
There is a bit of a chicken and egg thing when it comes to data, and over time if your app gains popularity you gain more data, and as you gain more data, the more accurate the algorithms should be. Sometimes you just need to wait for your app to reach a critical mass to really make it work correctly, especially if external data sources don't do the trick.
Regardless, we are at a point thanks to cloud computing where any developer with a modicum of ability has the resources to build artificial intelligence technologies into an applications. None of that means that every AI-fueled app is going to be successful, especially as everyone has equal access to the same tools, but it does mean that we are in age where we have some amazing technology within our reach — and we can in large part thank the cloud for that.
Photo Credit: A Healh Blog on Flickr. Used under CC by 2.0 license.