Feb 05 2020

What Government Can Learn from a New AI Playbook

Agencies can use ACT-IAC’s new guide to determine when to use artificial intelligence and how to effectively implement such solutions.

Federal agencies are starting to use artificial intelligence in a variety of ways to enhance their missions, including to increase cybersecurity, to process packages more efficiently and even to help battle diseases.

However, the government currently has an ad hoc process for developing and implementing AI projects, absent a forthcoming policy guidance from the Office of Management and Budget

Last week, the American Council for Technology and Industry Advisory Council (ACT-IAC) filled the void and released its AI Playbook for the U.S. Federal Government

The playbook runs through five different phases: assessment to determine if AI is the appropriate technology to solve a problem; readiness to prepare the agency for the use case; selection to enable the agency to operationalize the use case; implementation to put the solution into place; and integration of the solution. 

“I think — from a CIO standpoint — it gives you a roadmap to be able to implement a much smaller prototype to see success or a much bigger project to see success,” one of the playbook’s authors, Abeyon CEO Mallesh Murugesan, tells MeriTalk.

“AI has the potential to help government mitigate fraud, reduce errors and lower the cost of paper-intensive processes, while enabling collaboration across multiple divisions and agencies to provide more effective and efficient services to citizens,” ACT-IAC notes. “Moreover, the adoption of AI may also allow government agencies to provide new value-added services to citizens which can generate new sources of revenue and achieve agency objectives.”

VIDEO: What are the seldomly asked questions around emerging tech? 

How Agencies Can Effectively Implement AI

ACT-IAC hopes that agencies can use the playbook each time they plan to deploy an AI project, and that they can use it to figure out how to deliver a minimally viable product, proof of concept, pilot projects, initial operational capability and a fully operational system to support their overarching mission objectives

The playbook can provide “a step-by-step process to deliver solutions and provide a scalable product that is sustainable through the lifecycle of development, implementation, and recapitalization,” ACT-IAC notes. 

IT leaders interested in using AI will need to “define the appropriate stakeholders and the group (network peers) that will participate in the steps outlined in the playbook.” The playbook serves as “a catalyst to the necessary cultural transformation at the core of change management essential for the evolutions of today’s organizations.” 

The five phases included in the AI playbook include:

  1. Problem Assessment: Develop a vision and business objectives to see if AI solutions can address specific use cases while delivering results that optimize services and operational delivery.
  2. Organizational Readiness: “Engage AI subject matter experts and consider nuances that accompany an AI solution” through creating a project management office and establishing an AI-tailored business, functional and technical requirements, and implementation plans.
  3. Solution Selection: Investigate the business consideration, AI requirements, deployment models and procurement options that will establish an optimal provider selection.
  4. AI Implementation: Customize and configure AI solutions to meet operational objectives.
  5. AI Integration: Integrate AI solutions into the agency infrastructure. One of the playbook’s co-authors, Frederic de Vaulx, vice president of Prometheus Computing, tells MeriTalk that the playbook’s authors wanted the document to be able to guide IT leaders from one phase to another no matter the AI project they are working on. 

“When we developed the playbook, we really tried to make sure that it’s not just about the adoption of the technology and trying to figure out what the technology is but also how it impacts or is influenced by the rest of the organization,” de Vaulx says. “How is an agency going to integrate it within the rest of the organization? How is the workforce affected, or how will it will affect the technology?” 

The AI playbook is designed to help agencies get their internal IT teams collaborating on AI initiatives. “The CIOs, project managers and acquisition specialists are some of the groups that need to be a part of this journey in order for you to be successful,” de Vaulx says.

Design Cells/Getty Images