Page 35 - FCW, October 2017

EN TE RP RISE
SPONSORED REPORT
OUTL K
 Technology Modernization
The early adopters tend to have both the budgets and the in-house expertise to accomplish that, says Bob Osborn, Federal CTO at ServiceNow, and a former CIO for several federal agencies. “But the majority of agencies are not like that,” he says. “They provide citizen services and for them it’s a much tougher equation, since they likely don’t have the budgets to assess and integrate new technologies like AI.”
That provides an opening for companies such as ServiceNow, he believes, that can provide AI and machine learning capabilities as an integrated component of its overall
systems will have to be separately accredited. So far though there’s little understanding within the various accreditation organizations about how AI-based software operates.
Platforms with integrated AI capabilities, with algorithms written specifically for those platforms, will presumably have an easier change of accreditation. The AI output will be a part of the platforms’ overall output and therefore evaluated as a part of that platform’s accreditation.
Another likely challenge to more widespread AI adoption throughout the government is pushback from federal employees. They will see their
assessment programs.” So the human component is safe for now, at least at the SEC.
Despite some of the more fanciful stories about AI that have dominated parts of the recent public discourse, Osborn says AI is not a true sentient intelligence. Most likely, it never will be. It will change certain business processes because they’ll no longer need humans in the loop. “But
we’ll always need humans to make key decisions on mission critical activities,” he says.
It’s no longer a case of whether or not AI will affect government IT. It already helps to run government,
according to the Deloitte study.
And cognitive technologies “could eventually revolutionize every facet of government operations, from virtual desktops to applications that can govern large, shifting systems.” What government organizations
are now faced with is making the tough choices about how and
where to introduce these new
technologies, and integrating AI with its existing workforce.
ServiceNow, Inc. is an Enterprise Cloud computing company headquartered in Santa Clara, California. It was founded in 2004 by Fred Luddy. ServiceNow replaces unstructured work patterns with intelligent, automated workflows so IT, HR, security, customer service, and everyone in the enterprise can work faster and more efficiently.
For more information, please visit: servicenow.com
“Human interaction is required at all
stages of our risk assessment programs.” —Scott Baugess, acting director and acting chief economist
of the SEC’s Division of Economic and Risk Analysis (DERA)
platform. Agencies will ship data to the ServiceNow AI engine, and receive output customized to their needs.
ServiceNow’s AI may not be as robust as a standalone AI engine, says Osborn, but given that it’s fully integrated into the company’s platform, this kind of “practical
AI” is available today and is certainly of providing up to 80 percent of
an agency’s requirements. The company is looking at applications such as customer service, security, and human resources; as well as
IT modernization.
That kind of all-in-one service also
manages some of the regulatory objections that might stymie AI applications for IT modernization. Certainly with standalone AI,
the algorithms written for those
jobs affected by the technology and some fear they may be completely eliminated. However, it remains uncertain whether or not that will happen to any significant degree. Right now, the greater change is likely to be in the emphasis of certain skills.
At the SEC, for example, though machine learning is being used more and more, computers are not yet conducting compliance examinations on their own, says Baugess. It’s
not even close. “Machine learning algorithms may help our examiners by pointing them in the right direction in their identification of possible fraud or misconduct,” he says, “but machine learning algorithms can’t then prepare a referral or enforcement.”
Simply stated, “human interaction is required at all stages of our risk