AI in the Cloud Boosts Cloud Performance

For most enterprises, cloud-based AI services offer the easiest route to piloting and deploying artificial intelligence technology.

WEBINAR: On-Demand

No-Size-Fits-All! An Application-Down Approach for Your Cloud Transformation REGISTER >

Clearly, the combination of artificial intelligence and cloud computing will offer a critical advantage to those companies that leverage it. The challenge is, combining these two emerging technologies is easier said than done. 

First, to say that enterprises are embracing artificial intelligence (AI) would be a vast understatement. According to IDC, revenue for AI and cognitive systems will likely top $12.5 billion this year, representing a 59.3 percent increase from 2016. And the research firm forecasts that the market will increase 54.4 percent per year through 2020, at which point revenues will likely exceed $46 billion.

"Intelligent applications based on cognitive computing, artificial intelligence, and deep learning are the next wave of technology transforming how consumers and enterprises work, learn, and play," said David Schubmehl, an IDC research director. He added, "Cognitive/AI systems are quickly becoming a key part of IT infrastructure and all enterprises need to understand and plan for the adoption and use of these technologies in their organizations."

So far, it appears that enterprises are taking his advice. In July 2017 Vanson Bourne conducted a survey of 260 IT and business decision makers at enterprises with annual sales of $50 million of more. The resulting State of Artificial Intelligence for Enterprises report stated that 80 percent of organizations had already deployed some sort of artificial intelligence. Despite those initial investments, 30 percent of respondents believed that their companies had not invested enough in AI and would need to increase spending over the next three years to remain competitive.

But while enterprises might be enthusiastic about the potential of AI, they also recognize that successful deployment of the technology poses significant challenges. In that same survey, 91 percent of respondents anticipated barriers to AI adoption. More specifically, 40 percent didn't believe they had the right infrastructure to support AI. Also, 34 percent said they didn't have the right talent, and 33 percent said the technology is still unproven, while 30 percent said they lacked budget for implementation.

ai in the cloud

Data Source: Teradata State of Artificial Intelligence for Enterprises

In order to overcome these barriers, many enterprises are choosing to tap AI services via cloud computing. This decision offers a number of advantages versus deploying AI capabilities in their own data centers.

Benefits of AI in the Cloud

In many ways, the cloud and AI seem made for each other, and experts predict that artificial intelligence services will be the next big battleground for cloud vendors. For example, earlier this year, Canalys Research predicted that future growth in the cloud computing market "will be fueled by customers using the artificial intelligence (AI) platforms cloud service providers are building to develop new applications, processes, services and user experiences."

Few enterprises — even among the largest — can afford to do fundamental research on artificial intelligence. However, all of the major cloud vendors are investing heavily in artificial intelligence research and development. By utilizing these services, organizations are able to leverage advances in AI technology without having to bankroll the research themselves. In addition, utilizing cloud AI services offers a number of other benefits as well:

  • Access to Advanced Compute Infrastructure — Machine learning and neural networks require massive amounts of parallel processing power. To meet that need, AI applications must run on systems with advanced graphic processing units (GPUs). However, those systems can be very expensive — making them somewhat impractical for pilot projects. Using a cloud service allows enterprises to get the compute resources they need for AI initiatives while paying only for what they use. And if a pilot project doesn't pan out, they can easily shut it down without being saddled with expensive hardware they no longer need.
  • Scalability — When enterprises experience success with their initial AI efforts, they often want to expand those projects. The cloud makes it easy to scale those projects up or down as demand warrants. And the cloud model makes it easy for companies to broaden the use of the technology to additional departments and business units within the organization.
  • Ease of Use — Competition is fierce for developers and data scientists who understand artificial intelligence technology. As a result, salaries are very high for these professionals. Many organizations find that they can't find or can't afford AI talent, and developing AI skills among their existing workforce takes time. However, the major cloud vendors are meeting this need by rolling out AI services that simplify both the process of creating and training machine learning models and the process of adding speech, image recognition or natural language processing to applications. That, in turn, helps companies overcome any lack of internal AI talent.
  • Access to the Latest Technology — The Teradata survey respondents rightly pointed out that AI technology is still young and changing every day. Because of their AI research and development investments, the major cloud vendors are rolling out new AI capabilities on a regular basis. If an enterprise were to invest in AI hardware and software for its own data centers, they may find that their technology quickly becomes obsolete. But using the AI cloud services allows organizations to stay at the cutting-edge of advancements.
  • Low Costs— This article has already mentioned the financial advantages of cloud-based AI several times, but it bears repeating. The cloud computing model allows organizations to pay only for the computing resources that they are using for their AI application deployments. That eliminates the need for costly upfront capital expenses, allows organizations to convert their infrastructure costs to operational expenses, and often reduces the overall price tag for artificial intelligence projects.

Challenges of AI in the Cloud

Of course, using cloud-based artificial intelligence also presents some obstacles — most notably compliance and security.

Currently, one of the biggest reasons enterprises are investigating AI technology is to perform predictive analytics. They are using cloud services to build and train machine learning models that can help them gain valuable actionable insights.

But those models are only useful if they are fed large quantities of data. And in some cases, that data may be sensitive, valuable to potential cyberattackers and/or governed by regulations. In some industries, organizations may be prevented from processing or storing some kinds of data in the cloud. And some regions of the world also limit where certain types of data can reside geographically.

Thus, before deploying a cloud-based AI service, organizations need to make sure that they have adequate security in place to protect their valuable data and that they are meeting all of their compliance requirements. They cannot simply assume that the cloud provider has all the necessary security measures in place, as the enterprise shares responsibility for security.

In addition, understanding the various AI cloud offerings available can be very confusing. Cloud vendors use a lot of different terms — artificial intelligence, machine learning, deep learning, cognitive computing, neural networks and more — and different vendors sometimes use those terms to mean different things.

The best way to see if a particular AI cloud service is going to meet your needs is to give it a try. Most of the vendors make that easy with free or inexpensive trial periods. However, this trial process is going to require some time effort.

Cloud Based on AI

So far, this article has focused on infrastructure as a service (IaaS) or platform as a service (PaaS) offerings that allow enterprises to deploy artificial intelligence capabilities. But that's only half the story.

Many cloud vendors are also tapping artificial intelligence to make their existing IaaS, PaaS and software as a service (SaaS) offerings better.

For example, Amazon has a service called Macie that uses artificial intelligence to help secure data stored on its S3 cloud storage service. Oracle is incorporating AI technology into its database software to create an Autonomous Database that can manage and tune itself. Salesforce embeds its Einstein AI into all of its enterprise SaaS offerings. And many, many other SaaS vendors are incorporating AI capabilities into their solutions.

In the end, even companies that don't directly use a cloud-based AI service will likely end up utilizing some form of cloud AI because AI will be a part of the other cloud-based software they use. Gartner has predicted that "By 2020, AI technologies will be virtually pervasive in almost every new software product and service," and that seems particularly true for cloud-based enterprise software.

Cloud-Based AI Services

Companies that are ready to get started with cloud AI have a number of options available. All of the major cloud vendors have AI services in production, and a number of other companies have either announced or rolled out cloud AI services. The list below offers an overview of some of the most noteworthy cloud AI solutions:

Amazon Web Services

  • Lex — The same technology that underlies Amazon Alexa, Lex allows developers to create chatbots and conversational interfaces.
  • Polly — Polly is a text-to-speech service that uses deep learning to generate lifelike speech.
  • Rekognition — Using deep learning technology, Rekognition analyzes images, identifying objects, faces, scenes and celebrities, as well as flagging objectionable content.
  • Machine Learning — The same technology used by Amazon's own data scientists to guide its business decisions, the Machine Learning service offers tools for creating models and obtaining predictions.
  • Apache MXNet — This service simplifies the process of adding deep learning to applications through the Gluon interface.
  • TensorFlow — Although TensorFlow is a Google-created tool, you can also run this deep learning framework on AWS.
  • Deep Learning AMIs — The Deep Learning AMIs are pre-configured AI instances for deep learning with Amazon Linux or Ubuntu.

Microsoft Azure

  • Machine Learning — The Azure Machine Learning Service incorporates all the tools developers need to add AI capabilities to their applications, including the Machine Learning Studio development environment, Spark, Docker, Cognitive Toolkit, TensorFlow, Caffe and more
  • Bot Service — Still in preview, this service aims to simplify the process of creating a chatbot.
  • Cognitive Services — This collection of APIs allows developers to add vision, knowledge, language, speech and advanced search capabilities to their apps.

Google Cloud

  • Cloud Machine Learning Engine — Based on Google's TensorFlow framework, this service allows users to build and train models to generate predictions.
  • Natural Language API — This REST API can extract information, understand sentiment, parse intent and more.
  • Speech API — This speech recognition tool converts audio to text in more than 110 different languages.
  • Translation API — The Translation API quickly translates text from one language into another.
  • Vision API — Using machine learning technology, this service can analyze image content and classify it accordingly.

IBM Bluemix




Tags: cloud computing, artificial intelligence, AI in the cloud


0 Comments (click to add your comment)
Comment and Contribute

 


(Maximum characters: 1200). You have characters left.

 

 

IT Management Daily
Don't miss an article. Subscribe to our newsletter below.