1. What are these standards?
The “99s” are a family of standards that serve two functions: 1) establishing a conformance program for organizations who wish to certify that they practice ethical oversight for artificial intelligence (AI) systems, and 2) creating a framework to enable individuals to certify that they can perform AI ethics oversight for specific AI applications. By utilizing the existing ethically-aligned design standards developed by the IEEE, and incorporating best-practices from existing standards for quality and risk management, this family of standards is intended to complement the IEEE CertifAIEd program—both to support organizations wishing to certify their products under CertifAIEd, and to operationalize AI ethics oversight as a core business function.
While the primary sponsor for the “99s” is the Computer Society’s AI Standards Committee (C/AISC), the SSIT/SC (Society on Social Implications of Technology Standards Committee) is the co-sponsor of these projects.
2. Why are they important?
The “99s” are focused on one target: helping industry understand what it means to be an AI ethicist. Because so few academic programs currently exist to provide a unified, accredited solution for what an AI ethicist should or ought to be (i.e., degrees and certificates), there is no standard metric whereby C-Suite executives can cohesively establish AI ethics boards or oversight committees/teams.
After all, the continuous “churn” of personnel performing AI ethics oversight in large tech firms isn’t the result of a lack of effort or desire to ensure that AI is operating under ethically-aligned design paradigms. Without uniformity—whether from an academic degree program, or a standardized conformance program—it is simply not feasible to measure how “good” or “bad” an AI system may or how well it abides by ethically-aligned design principles and philosophies.
3. What is a real-world example or case study of how they might help?
An organization wants to show a regulatory authority that they conform to several of the IEEE’s ethically-aligned design standards—say, IEEE Std 2089-2021, IEEE Std 2089.1-2024, IEEE Std 7000-2021, IEEE Std 7001-2021, IEEE Std 7010-2020, and IEEE Std 7014-2024. Rather than auditing the organization’s technical and quality documentation several times against each standard, the regulatory authority could instead verify that the organization is certified under IEEE Std 7999 and proceed to verify conformance against 7999 alone.
Beyond saving an organization time and money, the “99s” provide a method that enables organizations to understand how the various ethically-aligned design standards operate synergistically—while also enabling Recommended Practices and Guides to serve a more critical role in supporting the “state-of-the-art” nature of the product or device. This approach therefore encourages organizations to adopt and comply with a greater range of standards they may have previously thought were too difficult to implement, as it gives them a greater level of insight into how IEEE’s ethically-aligned design standards for AI interoperate within targeted domains (e.g., eMedicine, self-driving vehicles, generative multimedia).
4. What stage are they at?
The “99s” are still in the beginning stages of development, as the respective PARs (Project Authorization Requests) were approved in September 2024. Because we wanted to avoid trying to take people’s attention away from the holiday season in December, we only started holding monthly meetings in January this year—with April being our third working group meeting overall.
5. What is the current geographical or disciplinary spread of your working members?
At a glance, the bulk of our 111 voting members are based in North America or Europe. We also have a good number of participants from Australia, with participants from South America, Africa, and Asia (including Oceania) making up a smaller share of our group. In terms of disciplinary spread, “governance” personnel are in the majority—though it’s safe to say that we represent a large number of industries (e.g., medicine, law, computer system architecture).
6. What type of people might be interested or well-suited for these standards groups?
While the intent of the “99s” is not to generate a significant number of new requirements and original content, we do have to review a hefty quantity of standards to synthesize individual requirements. As our work involves separating out the existing requirements for specific AI applications (e.g., eMedicine, self-driving vehicles, generative multimedia) to create certification requirements for individuals performing AI ethics oversight day-to-day, domain experts who are impacted by the IEEE’s 7000 series of standards—or who otherwise helped bring those standards into being—are likely to be well-suited for this group.
I would also say that individuals who enjoy regulatory science (as an academic discipline) are likely to be well-suited for our working group activities, particularly when it comes to figuring out how to resolve conflicting requirements or address precise gaps in AI ethics oversight for specific AI applications.
7. What triggered your own interest in this area?
Having participated in the development of P2863 (Recommended Practice for Organizational Governance of Artificial Intelligence), I was already firmly engaged in working towards a more robust system of AI governance. But between seeing the “churn” of AI ethics personnel, and the lack of academic interest in AI ethics, I figured it wouldn’t hurt to push for a standard (or series of standards) that operationalized AI ethics oversight in industry—especially since I’m still a Young Professional and uncertain where my career path will be after member elevation. After two long years of planning, development, and seriously talking with experts in the field, the “99s” were finally sent to the New Standards Committee—and the rest, as they say, is history.
8. Call to Action
There is certainly a long road ahead for the “99s”, and there will be plenty of opportunities to engage and contribute. Our work will start on developing P7999 before we branch out and develop the more specialized requirements of P7999.1 and P7999.2. The community can check our public site to find information on our upcoming meetings or approved subgroups, and are free to contact me at [email protected] (if they can’t get a hold of our IEEE SA program manager Christy Bahn at [email protected]) for specific inquiries.
