Federal government IT leaders agreed on Wednesday that workforce skills and data availability are two of the most important issues facing the government as it gears up to harness the benefits of machine learning and artificial intelligence in cybersecurity.
“You need the people with the right skills,” said Dean Souleles, Chief Technology Advisor to the Principal Deputy Director of National Intelligence, during a panel discussion at the Billington Cybersecurity Summit focused on building successful agency AI programs. “We need to invest in the skills of the workforce,” he said.
“We need people that understand the promise of the technologies,” and how to use them, he said. Agencies also need access to what he called the “foundation” of digital technologies including such gear as tensor and graphics processing units.
And just as important, Souleles said, “you need data and lots of it. . . Curated data, not the kind that is lying around in piles.”
Lt. Gen. John Shanahan, director of the Defense Department’s Joint Artificial Intelligence Center (JAIC), said that “talent, culture, and data” topped his list of necessary ingredients. He added that the third item on that list – data – is further complicated by cybersecurity concerns.
And for government agencies looking to jumpstart AI efforts, tech infrastructure also remains challenging, the general said. “We have to curate the data, and figure out how to use systems that were never built for AI,” he said.
As for employing AI to assist in cybersecurity, Gen. Shanahan said one of his biggest challenges is determining “what is the baseline of normal.” He said his organization has 24 suppliers of cybersecurity data, but they use varied data formats. As a result, he said, “we are working on a cyber data framework” aimed at standardizing data curation. “Then we can do some things with standardized data,” he said.
“We don’t have that answer now,” added Souleles regarding defining a “normal” baseline for cybersecurity.
Swami Sivasubramanian, Vice President at Amazon Web Services, advised it’s “best to start out small” with AI development, and be prepared to invest substantially in time and effort to get the best results. He said AI development efforts can be expected to last for years, adding, “it’s a journey of discipline . . . it takes time.”
Lynne Parker, Assistant Director of Artificial Intelligence in the White House’s Office of Science and Technology Policy, said the White House will be holding a summit on AI in government on Monday, Sept. 9, that will feature discussion of AI use cases including in disaster recovery efforts and medical research and development.