A New Year for Artificial Intelligence

Artificial intelligence will have a huge impact in 2019. Agencies that adopt the technology will be able to analyze large volumes of data quickly, automate repetitive processes, increase transparency and improve efficiency of operations. Learn how you can launch an AI strategy in the new year.
Individuals and organizations alike are setting goals and resolutions for the new year. It’s a great time for government entities to think about how they can improve their operational strategies with artificial intelligence (AI). In recent years, government agencies have increased efficiency and effectiveness by finding new and better ways to use their data. And big data and analytics are helping governments better serve citizens in a more cost-effective manner.
AI will be a big topic of conversation in the public sector in 2019 since it presents an opportunity for governments to take the next step in their analytics journey. AI benefits include analyzing large volumes of data quickly, automating repetitive processes, increasing transparency and improving efficiency of operations – all important goals for governments everywhere.
Tailor AI by agency need
Like any organization, it’s important that government agencies first define what AI means for them. Simply put, AI is the science of training systems to emulate human tasks through learning and automation.

With AI, machines can learn from experience, adjust to new inputs and accomplish specific tasks without manual intervention. However, the organization implementing the technology decides its level of sophistication and the degree to which manual intervention is required.
Government agencies are fundamentally unique from the private sector in that success is mostly measured by mission effectiveness rather than profit or loss. And the accomplishment of the mission or mandate has constant challenges around situational awareness, operational tempo and resource constraints. These are all areas where AI can have a meaningful impact by helping improve transparency and oversight.
Getting started
Where should government agencies start as they consider AI technology? Our experience supporting effective AI implementations in government has taught us that any AI strategy for government should include the following:
- AI leadership team. This team is responsible for defining both the business and technology directions, approving the systems approach, providing adequate resources, setting success criteria and ensuring adoption throughout the agency. This team should always include leadership from technical, administrative and operational teams across the organization.
- Pilot program. With AI, it is always important to start small. Define what a pilot program will entail, including a specific business group, pilot timeline, data to be analyzed, amount of transparency required, ethical guidelines and success requirements to move beyond the pilot phase.
- Data management. AI systems are only as good as the data they are fed. Implementing AI within an organization is a great opportunity to reexamine and address any issues with existing data management systems. This can include digitizing paper-based forms, improving authentication standards and adding data lineage to better monitor data access and control.
- Analytical tasks. It’s important to clearly articulate the task you expect an AI system to perform. This helps with managing expectations throughout the organization and defining how the technology will affect mission effectiveness. Due to the nature of government work, possible areas for initial consideration might be analyzing unstructured text or automating case management tasks.

- Human oversight. There must be strict accountability in government and that holds true when implementing AI systems. AI algorithms must not be just black boxes; they must be easy to interpret so that everyone understands what factors were used for calculating output. This can be incredibly useful in discovering and combating bias embedded within data, improving accuracy measures and creating procedural guidelines for human decision making.
- Business adoption. For any technology to have real impact in government, agencies must incorporate AI technology into their daily workflows and provide feedback on what’s useful and what’s not. The most impactful AI systems are ones with constant feedback from business users who are ingrained in daily operations and directly responsible for carrying out the organization’s mission or mandate. Close interaction between data science and operations teams will improve adoption rates, as well as the quality of the AI system being implemented.
- Success criteria. Every government program must have clearly defined success criteria, and innovation efforts around AI are no different. Success criteria should include data management, security guidelines, transparency requirements, algorithmic performance, model deployment, integration with legacy systems and business adoption.
- Growth plan. The pilot phase should be seen as operationalizing the cornerstone of AI within the agency. The pilot should have a clear growth plan where each step builds upon the one before it. The growth plan may include additional skills, hardware and software. When moving between different business groups, remaining cognizant of communication and collaboration between teams will ensure that technology can be used to bridge existing gaps in these areas across the entire agency.
When considering how to introduce AI into government organizations, agency mandate and mission effectiveness must remain central when implementing the technology. Delivering government services always has been and always will be a human-centric task. AI is transformative technology and implementing it across government organizations will improve efficiency, service quality and level of impact to its citizens.
Learn the SAS approach to AI and how government agencies can implement it into their analytics strategies. Read white paper: sas.com/ai-gov
By Mary Beth Moore, AI and Language Analytics Strategist, SAS