Good AI Starts With a Trained Workforce, Government Experts Say

Agencies’ digital transformation efforts in areas like artificial intelligence must also consider workforce needs, according to a panel of government technology experts. 

Speaking at an ATARC event on Thursday, the panelists asserted that it does not matter how good the data or AI is, if people do not know how to use it correctly or understand it. As a result, the panelists emphasized the need for data literacy, education and training.

“I can build the best AI model, but if I put it in the hands of my investigator, and if he has a ton of questions, then we just lost them,” Ben Joseph, chief data officer for the United States Postal Service Office of Inspector General, said. “Earlier this year, we actually punched out a small program in terms of data literacy…so we educate my workforce, investigators, auditors and everybody else, like ‘how do you interpret data?’”

“It’s almost like you have to right-size the AI education for the position or the role that the individual is playing in the lifecycle,” William Streilein, chief technology officer at the Department of Defense’s Office of the Chief Digital and Artificial Intelligence Officer, said. “And so, somebody who’s in acquisition, certainly they are capable of knowing all the details, but they don’t necessarily need to. They need to know enough and what’s relevant for their role.”

While data literacy and training is important, Joseph emphasized that it is also necessary to have people with different skill sets. 

“We don’t want to invest a ton of time on transforming everybody into data scientists,” Joseph said. “We need a mix of people like data analysts, data engineers, data scientists and people who can answer, communicate, change and all that.”

Meanwhile, the Department of Homeland Security is working on a program to gather existing top-tier experts in different areas from across the agency. 

“What we’re trying to do around the black belt program is find who the experts are in the DHS organization,” David Larrimore, chief technology officer for DHS, said.

The DHS program will evaluate three components: level of training, which may be a certification; the amount of personal or professional experience on a particular topic; and deliverables to prove one’s knowledge and expertise in a particular area. 

“Because there is no way that of the 350 or so acquisition programs going on right now, everyone has someone who could be considered an AI black belt,” Larrimore said. “But wouldn’t it be great if a black belt from CBP could go spend six months over a FEMA program to help them get on with this.”

Beyond cultivating expertise in advanced technologies, like AI, the experts noted that employees need to understand the value of not just any data, but quality data. 

“We have to get our data in order, because the data will supply you the fuel for the analytics,” Streilein said. “Teaching best practices related to data is probably the most important thing. We like to say, ‘no new bad.’ Data gets created all the time and it’s just so easy to create it and not put the right labels on it, not put it in the right place. We use the term VAULTIS—which is an acronym for visible, accessible, understandable, linked, trustworthy, interoperable and secure. So that’s a lot. But if you can make your data VAULTIS, then you are hopefully AI ready. That’s certainly a good bar to shoot for.”

According to Larrimore, the quality of data must be continually verified.. 

“We have to constantly question the data we’re looking at, and it’s only through working with components, with the data providers, with the data stewards,” Larrimore said. “Are we actually understanding where the rubber meets the road, the brass tacks, the bottom line, up front of what had been presented to us, right? And it’s not until those conversations happen, till everybody kind of comes to an agreement on what information actually provides value.”

Panelists also highlighted the importance of different parts of an agency working together and getting to understand how their portion impacts the larger agency mission. 

“Often, we don’t allow the acquisition workforce to be able to see the full results of their work,” Udaya Patnaik, chief innovation strategist for the office of information technology category at the General Services Administration’s Federal Acquisition Service, said. “You have to be able to know how the work that you are doing on particular acquisitions is being used in agencies and be able to connect all the dots that say, like, ‘Ooh, because we’ve made this action over here, the data impacts way down here.’”