Takeaways from the AI in Acquisition Forum

The AI Acquisition Forum assembled by ACT-IAC last week highlighted the importance of artificial intelligence in federal government acquisitions as well as the ethical, legal and security risks associated with the tools. The event hosted speakers and discussion panels to address issues ranging from AI and federal acquisition regulations, to how the government is buying AI and the use of generative AI in procurements.

Throughout the event, government guests shared AI use cases for the acquisition process, both those underway in implementation and those envisioned for inclusion in the federal procurement environment. For example, Michael Derrios, a Senior Procurement Executive (SPE) at the State Department, stated AI use case potentials include a speak to chat feature to search the FAR, a bot to assist in proposal compliance, AI to allow real-time visibility into IT spending, and something of further reach such as an AI tool to expedite RFP development. Jeffrey Koses, SPE at GSA, views AI contributing to wage-range work, cross-searching the FAR and agency FAR supplements on acquisition.gov, and labor categories. Karla Jackson, NASA’s SPE, shared additional AI use cases in procurement operations including supplier management, performance management, surveillance of contract functions, market research and predictive analytics.

An example of AI in procurement already underway was described by Andrea Branson, Deputy Assistant Secretary for Budget, Finance, Grants and Acquisitions at the Department of the Interior. According to Banson, Interior has worked for more than three years to stand up an AI solution to sit in the background and train from a system acquisition professionals use, with future goals to use the tool for market and predictive analyses.

Nonetheless, acquisition professionals must take into consideration the ethical and legal implications of AI in procurement, consider certain aspects of the FAR, as well as ensure data and privacy protection.

The following are additional takeaways from the event that stood out the most for me:

• The rate of AI expansion is unprecedented with the introduction of generative AI. Large language models and natural language processing take AI from the “basement to the boardroom.” (Rebecca Weaver, VP at CGI Federal)
• Agencies are relying more on Other Transaction Agreements (OTAs) and grants vs. traditional FAR-based contracts for AI acquisitions. (Karla Jackson)
• A separate contract vehicle for AI is not needed, procurement of AI can be found among existing contracts. (Jeffrey Koses)
• Agencies must ensure that data rights are defined, and the right regulations need to be in place by the federal government. (Karla Jackson)
• Best practices in emerging technology procurements include early engagement between industry and government to understand the challenge and end goal, releasing affordability statements to level the playing field, and holding discovery sessions between development phases (i.e. vendors can come in and ask questions and learn about the requirements). (Scott Simpson, Digital Transformation Lead at the Procurement Innovation Lab at DHS)
• Procurement officials must consider both short- and long-term effects of AI in acquisition, for example agencies may need additional staffing to respond to the potential increase of evaluations and offers submitted by industry using AI in proposals. (Karla Jackson)
• An increase in requirements more prone to open-standards data vs. siloed and vendor-locked in data will help the U.S. defeat China in the AI race. (John Ferry, President/CEO at Trenchant Analytics)

Final Thoughts: The overarching sentiment from the event of the federal government’s approach to AI acquisition is that early and increased engagement with industry is essential. Another consensus among acquisition officials is to begin using AI low-risk areas of procurement, such as market research and compliance checks. The acquisition workforce must also continue to take security and bias risk into consideration, prioritizing security standards (i.e. FedRAMP, ATO) to AI systems and defined data readiness and authorities. Lastly, the success of AI in the procurement process rests on transparency and clear communication of AI use must occur by both bidders and contracting offices.