By Brandon Bulldis, Federal Civilian Engineering Director at World Wide Technology

From chatbots to call centers, many federal chief artificial intelligence (AI) officers have begun exploring how virtual AI assistants can improve both the employee and citizen experience. When successfully implemented, these tools have proven effective in getting citizens the swift support they need and augmenting employees’ workloads while helping to attract and retain top talent.

However, for many agencies, AI assistants are still in the early phases of adoption. Whether agencies are starting their journey or ready for more advanced models, they must consider the benefits and challenges associated with the technology, as well as their agency’s requirements and intended use cases, before they develop and deploy solutions within their organization.

Benefits of Virtual AI Assistants

Virtual AI assistants can provide citizens with quick and accurate information and even connect them with the right point of contact to assist them, creating a more personalized experience. For example, call centers that leverage natural language processing can improve operators’ workflows and help constituents get answers to frequently asked questions without needing to phrase those questions using exact key terms, streamlining their experience with the agency.

Internally, virtual AI assistants present an opportunity to narrow the government skills gap. By introducing these tools into their workflows, federal agencies can improve their recruiting power by attracting younger talent interested in the technology. Additionally, these AI solutions create upskilling opportunities for the existing workforce, helping to narrow the talent gap from within making agencies internal knowledge bases more easily digestible for the workforce.

Virtual AI assistants can also help federal employees with daily tasks, saving them critical time and enhancing productivity. By leveraging AI assistants, these solutions can augment menial tasks, such as responding to frequently asked questions and troubleshooting IT issues, returning time to employees for work that advances mission outcomes. For example, the Department of Veterans Affairs (VA) is piloting a generative AI chat interface to help employees with basic administrative work, like drafting emails and summarizing meeting notes, with early results showing 72% of users agree the tool has made them more efficient.

Challenges to Consider

Although agencies are already seeing strong benefits associated with virtual AI assistants, there are challenges chief AI officers should consider before beginning the implementation process.

Agencies are investing billions of dollars in their AI integration strategies; however, most do not have the skilled workforce to successfully develop and deploy tools like virtual AI assistants. Although the federal government has made progress in upskilling and hiring skilled talent in recent years, there is still a ways to go before IT teams have the know-how and staffing capabilities to utilize AI technology to its fullest potential.

Ensuring AI tools have access to high-quality secure data is also top of mind for chief AI officers. Without a strong data foundation, agencies will struggle to develop and deploy AI solutions that maximize outcomes for citizens and employees alike. As agencies continue to develop their data strategies, they must ensure they create the data infrastructure needed to securely support AI assistants and scale these initiatives in the future.

Additionally, while AI bias and security are a concern across sectors, federal agencies are in a unique position given the sensitivity of their data. The data that virtual AI assistants are developed with can include bias, putting citizens at risk of discriminatory outcomes. The vulnerabilities associated with these tools also create a new threat vector for malicious actors, putting citizen data and critical infrastructure at risk.

Best Practices for Deploying Virtual AI Assistants

As chief AI officers continue to develop and evolve their AI strategies, they must consider their goals and approach to security and skills training to ensure the successful deployment of virtual AI assistants.

Zeroing In on Use Case and Infrastructure: As chief AI officers look to develop and deploy virtual AI assistants, they need to evaluate whether the tools they select or build are anchored around a purpose-driven use case. For citizen services, this includes ensuring the tools are user-friendly to maximize their use and benefits. A key part of this planning, agency leaders should determine early on if the use case will focus on employees or citizens as that will help define data sources, security boundaries, and AI guardrails. Based on the intended use case, they must then determine if they plan to develop their own large language model (LLM) or partner with industry to deploy a tailored solution for their agency.

Taking a Zero Trust Approach: Agencies’ virtual AI assistants should be secured using a comprehensive zero trust framework to prevent potential LLM data leaks. This should include establishing strict access controls and identity management that align with guidelines such as the National Institute of Standards and Technology (NIST) Artificial Intelligence Risk Management Framework. Additionally, agencies must ensure the data they are leveraging for these solutions is securely stored and accessed.

Upskilling the Tech Workforce: For agencies to maximize the impact of virtual AI assistants, they must ensure their workforce has the needed skills to successfully leverage the technology. Chief AI officers should pursue hands-on learning opportunities – including labs, research, and trainings – to help their workforce gain practical experience with the latest AI technology.

To harness the full potential of virtual AI assistants, chief AI officers must take a holistic approach to their development and deployment. By anchoring these tools around purpose-driven use cases, securing them using a zero trust framework, and ensuring teams have the skills they need to leverage the technology, agencies can improve the employee experience while accelerating citizen outcomes.