Businesses Are Accessing AI In The Cloud

The cloud has been a huge shift in technology for small and medium businesses over the last several years.  All indications are that it is an enduring trend.  Businesses access platform and infrastructure capabilities through major providers like Amazon’s AWS, Google Cloud Platform and Microsoft’s Azure.  They also use cloud-based SaaS applications–many of which run on those same major cloud platforms.

As these platforms increasingly add Artificial Intelligence (AI) or ‘deep learning’ capabilities that can be accessed by tenants, SMBs are increasingly able to access this leading-edge technology.

Machine learning requires huge amounts of computing resources and is technically demanding–far outside the realm of possibilities for SMBs (and even many large companies) and most SaaS companies to develop on their own.  But the resources of the cloud giants are making it increasingly accessible today and that trend seems very likely to continue. 

The model is to make these AI capabilities accessible through APIs to developers and other users in the cloud–so AI doesn’t require understanding neural networks technology and software, but rather just requires little more than signing up for an option so your software has access to the capabilities.

Last year, MIT’s technology review noted some of the new AI features Amazon had announced for AWS:

Chief among the new offerings are a series of off-the-shelf AI software packages that users can run on Amazon’s servers. There’s Transcribe, which converts speech in audio files into clean, time-stamped, punctuated text. Translate uses deep learning to shift texts between seven languages. Comprehend can detect sentiment in text. And Rekognition can detect and track people, activities, and objects in video. None of that is new technology, but the idea here is to make it easy for newbie developers to embrace AI on Amazon’s cloud.

Elsewhere, the firm has also launched an AI development platform called SageMaker, which is designed to make it easier for developers to build and train their own neural networks and run them on petabyte-scale data sets. And there’s also a new deep-learning-enabled programmable video camera, called DeepLens (pictured above), that is powerful enough to run its own AI algorithms and seems to be intended as a gateway for developers keen to understand what AI can do when combined with hardware.

All of this sends a very clear message from Amazon: it wants to make using AI on its cloud as easy as possible, like tossing a few files up on Dropbox (almost). It’s a savvy move, and the off-the-shelf algorithms in particular will no doubt prove popular with companies that want to start using AI but don’t quite know how. Will it be enough to fend off rivals? That remains to be seen.

Google Cloud Platform offers its own popular TensorFlow open-source AI software.  Microsoft and AWS have announced a partnership for a competitive AI software program, Gluon.

At the end of the day, there has been and continues to be a lot of hype about AI.  Machine learning does offer real capabilities today, but also suffers from limits–as anyone who has used Amazon’s Echo or Apple’s Siri knows.

That being said, discrete AI abilities like language translation, picture image recognition, language translation and text-to-voice capabilities (through Amazon’s Polly) are increasingly showing up in the resources of SMBs.  It’s an exciting time.