How Can Governments Use AI to Improve Procurement?

This post first appeared on IBM Business of Government. Read the original article.

Tuesday, July 19, 2022

By using artificial intelligence, agencies can improve their procurement processes.

** This blog was first published on The Regulatory Review website.

Perhaps no governmental process creates the perception of a complex bureaucracy more than the U.S. government’s procurement system.

Agencies follow the Federal Acquisition Regulation—or FAR, for short—or its U.S. Department of Defense counterpart, DFAR. The FAR consists of 37 chapters at over 2,000 pages, plus various agency supplements.

Within the context of this highly regulated procurement process, agencies have experimented over the last decade with using artificial intelligence (AI) to innovate and streamline how the U.S. government makes acquisitions. These efforts show promise and agencies will benefit from learning the lessons of past efforts at innovation, as well as envisioning future possibilities for improving procurement.

Such efforts become even more necessary given that the number of pages in the FAR does even not include agency-specific interpretations from chief acquisition officers, legal opinions that guide agency actions, protest rulings by the Government Accountability Office which agencies must account for to avoid future risk, and more.

And changing this stock of existing rules and guidance itself requires undergoing a complex process that, as with any regulatory action, must follow the Administrative Procedure Act’s provisions for proposing a rule change for public comment, reviewing comments, and finalizing the rule. Any significant new additions or changes to procurement rules must be reviewed by the Office of Management and Budget before proposal and final action, a process that also includes interagency review and can involve separate public comment.

Any time procurement officials in government set out to propose a change to the federal procurement rules, they must assess the change in the context of these processes. And any time an industry official seeks to bid on a contract or to comply with procurement regulations, these processes set the rules of the road for engagement with the government. Industry must also remain aware of and respond to agency enforcement actions, even though these are not released to the public in any consistent manner.

Government agencies have experimented with AI to improve the procurement process.  The IBM Center for the Business of Government has released four lessons learned from these innovative efforts, which can help guide agencies, companies, and researchers understand past precedents and envision future possibilities.

First, the foundation of AI involves developing capabilities for adapting AI to government functions. In a pioneering report, Kevin DeSouza—then with Arizona State University and now with the Queensland University of Technology in Australia—identified success factors for any application of AI by government, focusing on data literacy about AI among agency officials, training for knowledge on how to apply AI, and systems to manage risk in implementing AI systems. He identified a variety of opportunities for building a stronger foundation for AI deployment, including the following:

  • Investing in IT infrastructure
  • Using cloud computing and open-source options
  • Using agile approaches to procurement and governance
  • Transforming the federal workforce
  • Engaging with AI experts
  • Developing collaborations with academic partners
  • Redesigning workflows to improve efficiency
  • Enhancing cybersecurity capabilities
  • Crowdsourcing
  • Working collaboratively across agencies
  • Monitoring systems to identify unintended consequences
  • Deploying robust auditing measures

Agencies seeking to develop AI in their procurement process can pursue these opportunities.

Second, experimentation requires AI pilots to improve the procurement process. The IBM Center collaborated with the Partnership for Public Service in a report to explore examples of early applications of AI to improve the acquisition function. One important example in the report centered on the U.S. Air Force’s experimentation with an AI system, designed to help acquisition professionals make sense of complex acquisition regulations and speed the process of buying goods and services.

The Air Force’s pilot project involved uploading thousands of regulations, contract cases, acquisition training material and Defense Department policy to a database. AI technology then helped answer queries from federal contract officials and contractors about acquisition rules and regulations, such as how to proceed with a contract, what procedures to follow, and what contract a small business could bid on.

Although this work did not scale to operational capability, the experience demonstrated the power of AI to help government cut through decades of accumulated burdens in implementing procurements.

Third, engagement means government working with industry to evolve applications. The Procurement Innovation Laboratory (PIL) at the U.S. Department of Homeland Security has established a center of excellence for procurement innovation, including AI and other emerging technologies among its capabilities.

The IBM Center and the Partnership for Public Service have reported on the PIL model and its process for engaging industry to apply new AI applications for improving the acquisition process. The PIL exemplifies the value of public-private partnerships and helped capitalize on engagement with industry, rather than being limited by the assumption that such interactions cannot be accommodated.

In addition, the PIL adopted a user orientation—adopting the perspective of those who use the procurement system and how they interact with it. The PIL has incorporated elements of an agile approach, iterating innovation in short phases with learning built in. Support from the top agency leadership has helped facilitate innovation and risk-taking.

Finally, maturity occurs with advancing AI capabilities across the acquisition organization. A recent IBM Center report, also authored by Kevin DeSouza, sets out an AI Maturity Model for public sector enterprises—advancing from ad hoc experiments to enterprise-wide transformation. The model includes incorporating AI best practices into acquisition operations at higher levels, such as through agency-wide policies on AI acquisition and the development of a robust ecosystem of external partners that can generate opportunities. Another key best practice would help an agency to put in place a formal process for learning and improving its AI acquisition, deployment, and maintenance.

A similar transformation model that I and other leaders in government and industry developed several years ago can guide acquisition organizations in maturing to leverage innovation like the application of AI to their craft.

This model—Acquisition of the Futureidentifies levels of growth for procurement organizations, with a special focus on innovation and reform of longstanding bureaucratic processes. This model is currently housed with the government industry collaborative American Council for Technology-Industry Advisory Council.

As these and similar initiatives demonstrate, leveraging AI to develop procurement innovations can help agencies and companies work together to develop and implement acquisition strategies that clarify requirements and identify best value bidders. They can cut through the seemingly endless provisions of existing policy and guidance and enable rapid action to meet agency needs.

To help scale and sustain this evolution, hiring and training skilled acquisition professionals will be needed, and initiatives such as the federal Digital IT Acquisition Professional Training program can help.

Overall, innovations in procurement processes can complement programs to acquire AI. The key to a better procurement system involves developing a mutually reinforcing pathway for government and industry to advance best practice and public value.

Leave a Reply

Your email address will not be published. Required fields are marked *