PROJECT BRIEF

BACKGROUND

Procurement is the process by which governments and state-owned enterprises acquire goods, services, and works necessary for public service delivery. While it remains a cornerstone of municipal operations and fiscal stewardship, the increasing adoption of Artificial Intelligence (AI) tools introduces new challenges related to privacy, discrimination, and accountability. Procurement staff, therefore, require clear methods for identifying when AI tools pose a sufficient risk that necessitates escalation for further technical and policy review.

AIMS

The City of Tempe, Arizona currently lacks a structured procurement review framework to guide non-technical staff in evaluating and flagging AI-specific risks while reviewing Tempe’s AI vendor questionnaire during the procurement process. This project aimed to develop a non-technical review framework that included a list of prevalent defined AI procurement risks, along with their causes, impacts, and mitigation procedures. Additionally, it analyzed current peer city, state, and international approaches to addressing AI-related procurement risks and solutions. Building on Tempe's current procurement process, the team provided guidance for procurement staff on risk evaluation and improvements to their current vendor questionnaire. 

METHODOLOGY

The team conducted a literature review of frameworks from the National Institute of Standards and Technology, the United Nations Educational, Scientific, and Cultural Organization, and the Organisation for Economic Co-operation and Development to inform recommendations for improving Tempe's AI vendor questionnaire and operational practices. The team also conducted a Latent Content Analysis to identify insights that procurement staff should extract from vendor responses to inform their decision-making. The team consulted existing research on AI risks to inform the development of the risk taxonomy. Furthermore, the team conducted interviews with procurement staff and researchers to refine framework content and structure.

DELIVERABLES

Deliverables included a non-technical framework that includes a (1) list of defined AI risks, (2) analysis of existing approaches to addressing AI-related procurement risks, and (3) guidance for procurement staff on evaluating risk levels and improving the current vendor questionnaire. 

IMPACT AND FUTURE WORK

This framework enhances Tempe's procurement process by guiding staff through structured evaluation of AI-related risks and clarifying when tools warrant further review. Analysis also provides guidance on assessing risks associated with each criterion and their potential impact.