Agencies Prioritize Pilots, Literacy to Overcome AI Challenges

The Department of Labor (DOL) and the Department of Housing and Urban Development (HUD) are developing change management strategies and leveraging sandbox environments to test AI and boost workforce literacy to ultimately drive operational improvements, AI experts said during GovCIO Media & Research’s AI Summit Nov. 7.

Around five years ago, DOL moved all its information technology, 27 agencies and workforce to a centralized Office of the Chief Information Officer (OCIO) organization, adopting a shared services model, which created an existing establishment for DOL to lean on as it continues to explore AI.

“All of the partnerships with agencies, all of the knowledge of the mission applications, that helps us. That centralized service model helps us a great deal to tag on AI onto that,” said Mangala Kuppa, chief AI officer at Labor. “Engaging the whole of the organization with the footprint we have, with a centralized approach, with partnerships, with our agencies, is how we are going about it.”

Over the past six months, Kuppa said adoption is faster than what she anticipated. She attributed this success to the centralized approach, engaging the entire organization in adopting and understanding AI. DOL is being intentional about educating its staff to drive literacy and reduce anxiety around AI adoption, calling out that AI is meant to enrich jobs, not take them away.

HUD is taking a similar approach to ensure its workforce is “utilizing what’s available and understanding what’s possible,” said September Secrist, the agency’s chief of staff of the CAIO and CFO.

While there is a great deal of excitement and buzz around AI, the technology doesn’t come without challenges, especially in government with the red tape surrounding tech procurement and integration. Both agencies face two primary challenges: resources and data.

“Evidence-based decision making is a big deal in government. Not having data to go by when you’re making decisions, [wondering] should I jump right now and invest in a tool or should I make it’s a judgment call, that piece is still a huge hurdle,” Kuppa said.

Secrist noted that President Biden’s AI memo is an unfunded mandate, so HUD is facing challenges finding the resources to respond to the memo’s calls. Additionally, HUD is working with outdated systems that aren’t interoperable.

“We’ve got 50 years of different technology stacks that don’t communicate, that don’t talk to each other, the data is not necessarily clean for interoperability, and that has been a big challenge that we found: how do we how do we make sure the data is clean and accurate in order to use AI tools to make sense of it, to read it, to have some of those integrations?” she said.

As agencies battle funding challenges and continue to modernize to meet the needs of emerging technology, leaders are testing pilots and creating sandbox environments to drive organic growth and adoption of AI, prove value and get the workforce on board with the tech.

“Doing nothing is just as damaging as doing something with AI … So, we must do something. We must adapt AI. We must be responsible with it,” Kuppa said. “We must create — beyond trainings, beyond literacy — sandboxes for people to try [AI], even if you have to put some investment into it, so that they can see the value and resounding successes … creating those channels within the organizations for people to share what they’re learning [is the next step].”

DOL is also creating community-of-interest forums and working groups to share lessons learned. As part of this plan, DOL is also identifying tech enthusiasts to champion these forums and working groups. These AI champions are rallying excitement around AI adoption, increasing education of the tools and driving adoption.

Communication is key, Secrist added. HUD is focused on breaking down silos, ensuring AI isn’t a “techy solution,” that “AI isn’t happening to [the workforce],” but instead, a tool that is easily used and understood by the entire enterprise.

Half of HUD’s workforce will be eligible for retirement in five years, so the agency is facing trouble with AI adoption. HUD is developing a communication strategy to demystify AI and ease anxieties around the emerging tech to ultimately drive adoption.

“We’ve been trying to communicate, educate and encourage. We’ve got some centralized training that we’ve been asking team members to take, some self-assessments, to see where you are in your understanding of AI and help us understand what additional trainings would be helpful for you. It’s an ongoing conversation … [to] bring them along and really empower them to envision how they want to solve their work problems with AI,” Secrist said.

Government continues to compete with industry for top tech talent, but federal leaders are focused on organizational change management to upskill the existing federal workforce. DOL said 80% of the CAIO’s role is getting people comfortable with AI and being intentional about agency leadership to educate the workforce and drive adoption. The agency released an AI compliance plan in September to provide a roadmap for its AI journey.

“We are extremely intentional. It is more change management than anything else. It’s not just a technology. You must find a way to bring everyone together. You should not do it alone, and you cannot leave anyone behind. You have to say, ‘we have to do something here.’ Not doing is not an option. Not doing is riskier than taking the leap of faith and doing something with AI and hopefully it will improve your services,” Kuppa said.