Less than 10% … that’s how many artificial-intelligence test projects are estimated to be deployed into full-scale production in enterprise environments, according to a recent report from the International Institute for Analytics.

There are a number of reasons for this surprisingly small amount, including an overwhelming amount of data and the lack of easy-to-use tools to analyze it. It’s a problem that calls for operationalizing AI and machine learning, making it accessible and repeatable consistently.

“Ultimately, if you want to get business value from those models and all of the hard work that you’ve done, it has to be injected into the business process,” said Anant Chintamaneni (pictured), vice president and general manager of BlueData at Hewlett Packard Enterprise Co. “Operationalization of machine learning is ultimately the key, and that’s the progression that enterprises have to make.”

Chintamaneni spoke with Peter Burris (@plburris), co-host of theCUBE, SiliconANGLE Media’s mobile livestreaming studio, at theCUBE’s studio in Palo Alto, California. Burris was joined for a digital community event by co-host Stu Miniman (@stu), and they also interviewed Nanda Vijaydev, distinguished technologist and lead data scientist at HPE; Patrick Osborne, vice president and general manager of big data, analytics, and scale-out data platforms at HPE; and Wikibon analyst James Kobielus (@jameskobielus). Discussion focused on the roles needed in a data science team, HPE’s introduction of a new software solution to guide enterprises in analytics deployment, and the importance of operationalizing AI and machine learning in the business process.  (* Disclosure below.)

DevOps for data science

One issue with operationalizing AI and machine learning in the enterprise is that it takes a village. There are data scientists to select the right algorithms, data engineers to clean inputted information, machine learning architects to build predictors, and programmers.

These are all part of a data science team that drives what has become known as data-center operations. “At Wikibon, we refer to it as DevOps for data science,” Kobielus said.

To help this diverse team ensure consistency in practice, HPE introduced ML Ops in September, a container-centric software solution that is based on technology the company gained when it acquired BlueData last year. BlueData’s software platform combines containers, virtualization, and big-data tools to offer customers a better machine-learning and AI experience.

“ML Ops helps you scale from a data scientist or data engineer developing an algorithm on their laptop to be able to run that at scale in the data center,” Osborne said. “We have a number of high-profile and new relationships that we’re building for this new ecosystem around AI and machine learning.”

An architected approach

To build data science into the business process, organizations will need to take an architected approach. Deploying complex analytics tools into equally complex environments has hindered the ability to achieve consistent results. This places more emphasis on making tools easier to use and accessible for targeted users while maintaining a secure operationalized environment.

“This is a living ecosystem,” Vijaydev said. “From an enterprise point of view, it doesn’t matter where a model gets built. It does matter where users access it; it does matter where security is applied.”

Global spending on AI systems alone will reach $98 billion in four years, according to data released by IDC Research Inc. That accounts for two-and-a-half times the $37 billion that will be spent this year by enterprise customers.

“They’ve made lots of investment in talent and tools to create these models, but they have to figure out how to operationalize them,” Chintamaneni said.

Watch the complete video interview below, and be sure to check out more of SiliconANGLE’s and theCUBE’s CUBE Conversations. (* Disclosure: Hewlett Packard Enterprise Co. sponsored this segment of theCUBE. Neither HPE nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE