Artificial intelligence (AI) and automated decision systems (ADS) are increasingly being used to aid or replace government decisions and judgments that impact opportunities, access, liberties, rights, and/or the safety of their citizenry. Government entities have been consumers and stewards of public use technology for over half a century. But traditional government procurement models are no longer fit for purpose in the face of advanced technological change with exponential social impact.
AI and ADS, such as data analytics, surveillance technologies, and database management systems, present new challenges and concerns for the public agencies procuring this advanced technology and for the communities at large especially where procurement laws or policies fail to provide risk averse approaches to AI public procurement.
The Need for Standardization in AI Procurement
To date, there are no globally compatible or interoperable standards establishing a uniform set of definitions or a process model by which government entities can address socio-technical considerations throughout their AI procurement commissions. On the 23rd of September 2021, the IEEE SA Standards Board approved a new project and Working Group to address such a need in the development of international standards.
The new Project is IEEE P3119™ Standard for the Procurement of Artificial Intelligence and Automated Decision Systems which aims to address the needs of government workers, policymakers, and technologists to make meaningful and accountable choices that are transparent about the socio-technical considerations and impact of AI products, services, and or systems on the public.
As a key member of the IEEE 7000™-2021 Standard Model Process for Addressing Ethical Concerns during System Design which was recently launched, I was contacted to consider being Chair of the IEEE P3119 Working Group and to help create the PAR (Project Authorization Request) that would establish the scope for this new and unique project. I felt honored and ready to accept the role.
For both public and private sector organizations, IEEE 7000-2021 integrates ethical and functional requirements in systems engineering design and development in order to mitigate risk and increase innovation. Similarly, in IEEE P3119, government entities and their private sector partners will benefit from a base standard that aligns best procurement practices with the critical evaluation of AI and ADS to mitigate risks introduced by AI/ADS in the communities they serve. Like IEEE 7000-2021, the new standard’s process model for IEEE P3119 will also consider the IEEE Ethically Aligned Design General Principles as imperatives to support innovation in the public procurement of AI.
Process models and requirements using a uniform set of definitions have informed countless public and private sector problems on the planet by providing consistency, documentation, controls, and standardization of best practices. Their technical value to humanity is their generalizability across domains, levels, and geographies.
The IEEE P3119 Working Group intends to establish a uniform set of definitions and a process model for the procurement of AI and ADS by which government entities can address socio-technical and responsible innovation considerations to serve the public interest.
The process requirements include a framing of procurement from an IEEE Ethically Aligned Design (EAD) foundation and a participatory approach that redefines traditional stages of procurement as: problem definition, planning, solicitation, critical evaluation of technology solutions (e.g., impact assessments), and contract execution.
The scope of the standard not only addresses the procurement of AI in general, but also government in-house development and hybrid public-private development of AI and ADS as an extension of internal government procurement practices.
The purpose of IEEE P3119 is:
- To establish a uniform set of definitions and process requirements that address the socio-technical and responsible innovation challenges in the procurement of AI/ADS,
- To help support agencies adapt their processes for procuring AI systems responsibly for the public interest, and
- To promote ethically aligned values and robust public engagement in the process model.
IEEE P3119 stakeholders will include government entities, community advocates, legal, data science, and social science experts, vendors (AI & ADS vendors and service providers, platform developers), public and private sector research foundations supporting public procurement of technology, philanthropic organizations, procurement experts, and the public at large.
The Future of Consensus Building and How to Get Engaged
Similar to laws, technical and socio-technical global standards take years to write and get approved by their authorizing institutions when consensus building is the core method. The starting roster for IEEE-SA’s P3119 differentiates itself from others due to its size and variety. A majority of the forty-eight individuals interested in the P3119 Standard Working Group show a robust engagement from a North American cohort. But many other global stakeholders from twelve countries thus far (Argentina, Belgium, Canada, Chile, China, Finland, Germany, Netherlands, Nigeria, Spain, UK, and the USA) have also joined the roster to express interest in writing the Standard for the Procurement of Artificial Intelligence and Automated Decision Systems.
I will be leading an incredibly diverse group of cognitive, social, and technical subject matter experts from a wide variety of disciplines including: social and data science, law and government, software engineering, computer science, information technology, geographic information systems, community advocacy, innovation research, sociology, technology policy, anthropology, and many more.
The Working Group’s first three activities will be:
- TERMINOLOGY STRATEGY: Build a consensus on the strategic approach to the P3119 uniform set of definitions and decide the analytical primacy of the same. A critical evaluation of a global database of AI and ADS procurement definitions is one of our first outputs.
- GLOBAL LANDSCAPING: Learn from each other what the world knows about AI and ADS procurement and what countries have done in the past (critical evaluation of use cases) to address the challenges. A critical evaluation of a global database of use cases is another early output.
- PROCESS MODEL AND REQUIREMENTS STRATEGY: Build consensus on what type of process model and process requirements will be sufficient to encourage responsible innovation in AI/ADS procurement within and between government agencies across global jurisdictions. An architecture for processes, products, and people will form a straw man structure for the process model and its requirements.
Because AI systems and their data provenance are rarely finite local solutions and because AI has exponential impact factors that can affect millions of people in multiple communities simultaneously, it’s imperative that the process model and process requirements critically evaluate these advanced, often intercontinental features of AI.
- “How will AI solutions out there impact us right here?”
- “What impact does X solution have on the present and possible future human rights of vulnerable community populations?”
- “How can we engage citizens in the critical evaluation of AI solutions during AI procurement commissions?”
- “How are countries innovating AI procurement in responsible and ethical ways that encourage robust participation from their communities?”
These are the types of questions that the Working Group will address to begin understanding the functions of a responsible and ethical AI procurement process model. They are very eager to begin the many challenging conversations about how to provide the world with a new international standard on AI procurement. Their first Kick-Off Working Group Meeting was on October 21st, 2021, at 11:00 am to 1:00 pm U.S. Central Standard Time.
There is also an open Call for Participation for anyone who would like to join the Working Group. Participation is on a volunteer basis and is free and open for all to join.
Author: Gisele Waters, Ph.D. IEEE P3119 AI Procurement Working Group Chair; Director of Service Development and Operations at Design Run Group