An intelligent cloud?

Today's new cloud-only offerings are trainable and open up a whole new consideration for the role of the cloud

Connected and Intelligent Cloud APIs

For the most part, many of the current cloud offerings have previous frameworks that could run in a local infrastructure as well. New cloud offerings are what I would call "cloud only" -- a series of tools that only exist in the cloud.

Here is a sample of Microsoft Azure's new offerings. Notice, unlike previous announcement concerning virtualization, databases and operating systems, these all have a different ring to them.

  • Emotion API -- Personalize experiences with emotion recognition
  • Language understanding Intelligent Service API -- Teach your applications to understand commands
  • Text Analytics API -- Perform text to sentiment analysis, extract key phrases and detect topic and languages
  • Face API --detect, identify organize and tag faces in photos
  • Speech API -- convert speech to text and back again and understand its intent
  • Recommendations API --Predict which products your customers are most interested in based on their previous transactions

What makes this all possible is that large cloud providers of non-hosting services, such as Search, are opening their internal engines, which have done billions of queries and millions of hours of analysis, to the public. Google and Microsoft have invested billions of dollars into understanding their data in meaningful ways. More importantly, they have existing revenue-generating business based on these abilities so they continue to invest heavily. Mix that with a hefty dose of competition and fear and you can see how we got to where we are today: the intelligent cloud.

These intelligent services will offer abilities to companies of all sizes that don't have the means to build such expensive and massive systems. Even if these companies could build similar systems, these services only work when millions of people continually use them. The systems have to be trained; they can't be programmed. Training is complex, takes lots of time, and requires lots of processing power.

There are several aspects of new features like Azure's that go beyond the technology:

Applications will get smarter. Of all the changes this is most exciting to me. Apps are mostly dumb, lacking any real ability to do what you want without explicit directions given in a way that computers understand. The model will start to invert -- instead of humans constantly trying to figure out what computers need to perform a desired task, computers will figure out what humans need to perform a task.

Programming styles will change. Most of today's programming is really just a complex form of "IF THIS THEN THAT." With intelligent systems, programming will need to incorporate a more statistical approach. I have never built a system that performs a function based on something like, "When John is happier, suggest a task from his to-do list that he likes to do when he is happy." 

Testing I feel really bad for testers already. It was hard enough to test the convoluted and complex systems we have today, so testing systems of the future will require a different set of tools. Developers who work with artificial intelligence have a taste of this already. Anyone who has ever trained a computer system knows that testing these types of systems takes a different skill set. Telling if a system is actually working or not takes time.

Roles will change, or the concept of roles will change. I am not sure if these new tools will simply create a whole new set of roles like Web and mobile did, but remember, web designers only existed after the web grew in popularity. There could be a title called Senior Facial Programmer like we have Senior Mobile Developer. There may be a wave of cloud titles that get created, but the rate and diversity of these types of systems leads me to believe that the traditional developer, tester, manager, architect paradigm will not hold up much longer.

It will take time for new services like this to be adopted, but if recent examples such as the adoption of NoSQL databases, which technically aren't SQL or Databases, are a guide, we will be seeing a lot more of these services integrated into our systems within 10 years.

This article is published as part of the IDG Contributor Network. Want to Join?

From CIO: 8 Free Online Courses to Grow Your Tech Skills
View Comments
Join the discussion
Be the first to comment on this article. Our Commenting Policies