DALLAS – AT&T and Tech Mahindra are developing an open source artificial intelligence (AI) and machine learning platform that will be hosted within the Linux Foundation. The Acumos Project is expected to be launched early next year.
The project is designed to provide a common framework and platform to assist in the building and managing of AI platforms. The Acumos Project will also allow developers to connect individual applications to create AI and machine learning products.
The project will provide editing, integration, composition, packaging, training, and the deployment of AI microservices regardless of language used to develop those microservices. Those microservices can also be linked together to create more complex and sophisticated AI services.
Basically, organizations should be able to take any AI and machine learning application, move them into the Acumos environment, and come out with a production ready service.
Linux Foundation executive director Jim Zemlin said he expects the project to target the autonomous vehicle, drone, content curation, and analytics markets.
At the AT&T: The Summit conference in Dallas this week, Chris Rice, senior vice president of advanced solutions at AT&T, said a lack of clarity in terms of moving AI beyond the developer community and into a production environment was behind the company’s work on the project. Rice explained that AT&T viewed the industry as needing a way to make AI application development a repeatable and accessible process and “lower the bar of entry so that those without the deeper language knowledge can work in the space.”
Rice said AT&T felt it was in a position to help propel this work because of its deep history in data analytics. He noted the company has been using those tools internally but had not moved to push them out to a broader ecosystem.
“We wanted to make a difference and to provide leadership but not be overbearing,” Rice said of AT&T’s involvement with the Acumos Project. “I think we have the right combination to provide that leadership and also the humility that goes with it.”
During a keynote address at this week’s AT&T event, Gartner Research Executive Vice President Peter Sondergaard said the firm expects AI to create $2.9 trillion in business value and 6.2 billion hours of work productivity in 2021.
“In 2021, AI will be a net job creator,” Soundergaard predicts.
During his own keynote address, AT&T CEO John Donovan said the company built on its past software work for the new AI project.
“This software foundation allowed us to launch Acumos,” Donovan said.
Rice provided more color by noting AT&T was using the lessons it learned in open sourcing its ECOMP framework into the Linux Foundation. ECOMP eventually morphed with the Linux Foundation’s Open-O Project to form the current Open Network Automaton Platform (ONAP).
Rice noted that ONAP was really a way to extend software-defined networking (SDN) across telecommunication infrastructure. He said the project has shown great progress since it was launched earlier this year, and it showed AT&T’s ability to work within the Linux Foundation’s framework.
While similar in terms of their development structure, Rice did note that ONAP and Acumos were not required to run each other. But, the use of open application programming interfaces (APIs) would allow for easy integration if desired.
“It’s not a strong tie that requires one to be there for the other to work, but for the network use cases they are more implementable in ONAP,” Rice said. “Acumos is just a natural way to move AI into SDN more broadly.”
He said the use of the two together would allow for the closed loop automation within the ONAP framework to “change” using the Acumos AI and machine learning as it “got smarter.”
“And, we are showing that this is something that can be done in open source,” Rice added.