RTI uses cookies to offer you the best experience online. By clicking “accept” on this website, you opt in and you agree to the use of cookies. If you would like to know more about how RTI uses cookies and how to manage them please view our Privacy Policy here. You can “opt out” or change your mind by visiting: http://optout.aboutads.info/. Click “accept” to agree.

Insights

Government Agencies Embrace AI Adoption Strategies at AI World Government 2019

The growing impact of artificial intelligence (AI) is being felt in all major industries and government agencies are looking to take advantage of AI as well. Due to the different levels of AI maturity and readiness that exist across the government, there is value in understanding the many stages of AI solution development.

The AI World Government 2019 conference had presentations ranging from identifying the right AI projects to begin with to determining the infrastructure needs for organizations with existing AI use cases. Here were the top takeaways that I had from the conference:

  • Create a data strategy before an AI strategy. It’s not possible to utilize AI without TONS of data work first, so start there, with AI in mind. In this context, “data work” means locating, cleaning, and preparing data for use in AI algorithms. It also includes addressing issues such as data governance, privacy, and stability of data sources over time. These are non-trivial tasks, and government agencies have it right by first creating Chief Data Officers in each agency (as opposed to starting with Chief AI Officers). In some cases, it can make more sense to first identify a problem and then find your data and get started, however in this case, don’t expect your data to be mature. There is no avoiding the detailed data work that is required to move from a small R&D proof-of-concept to a large-scale production solution.

 

  • All AI solutions are bespoke, meaning that today's AI solutions are directly tied to the data that they rely on (imperfectly quoting Brian Thomas of NASA). Significant updates are needed if changes are made either to the underlying data or the target application. While some platforms can host multiple AI solutions, those AI solutions themselves don't handle multiple applications easily.

 

  • Enterprise-level code and artifact repositories are the best way to prevent re-engineering and encourage re-use, specifically for shared data sources. For example, Github is the most popular repository that underlies most open-source software. Enterprise versions of Github are available to establish within your company or agency to assist in the sharing of knowledge and code between your teams.

 

In fact, the above takeaways scale nicely, whether you’re considering AI adoption at a huge government agency, or for a small project in your department. Additionally, while getting data organized for an AI-oriented project, robotic process automation (RPA) opportunities will arise that will help improve overall data handling and data lineage. RPA improvements alone can sometimes be huge game changers before AI even gets off the ground.

Regardless of the maturity of your AI journey, I believe there are significant learning opportunities at conferences such as AI World Government. Level setting goals to other similar organizations and getting insight as to what might come next in your journey are important parts of the planning process.

Disclaimer: This piece was written by Edward A. Preble (Senior Research Data Scientist) to share perspectives on a topic of interest. Expression of opinions within are those of the author or authors.