How can organizations get started?
To effectively leverage the NIST AI RMF, organizations should begin by assessing their current AI capabilities and strategy, as well as how it intersects with broader ERM efforts. The framework is intended to be flexible, helping enable organizations to align practices with applicable laws, regulations and norms which may differ by industry or sector.
Once an effective baseline is established, organizations can start to apply framework insights on measuring risks, risk tolerance, risk prioritization and integration of risk management concepts related to AI.
As organizations’ AI capabilities mature, the NIST framework and its core functions should be revisited and supporting risk management capabilities should continue to be iterated upon to strengthen trustworthy AI.
How the essential characteristics of the NIST AI RMF align with Deloitte’s Trustworthy AI FrameworkTM
As AI and other advanced automated systems are becoming increasingly common tools used by organizations, Deloitte recognized the need to approach these evolving technologies in an ethical and responsible manner. As pictured in figure 4, Deloitte’s Trustworthy AI Framework has empowered organizations to build trustworthy AI and helped prepare them for the growing regulatory focus on AI and other automated systems.
The NIST AI RMF outlines seven characteristics for achieving responsible use of AI systems and effectively managing AI risk: valid and reliable, accountable and transparent, safe, secure and resilient, explainable and interpretable, privacy-enhanced and fair. Each characteristic builds upon the socio-technical viewpoint NIST advocates when implementing and managing AI technologies, however the accountability and transparency attributes also pertain to the external processes and context surrounding the AI systems.
Over the past decade, Deloitte developed its Trustworthy AI Framework based on hands on experience and cross industry leading practices to help clients throughout the AI lifecycle manage AI risk. Deloitte’s Trustworthy AI Framework is comprised of six characteristics: fair and impartial, robust and reliable, privacy, safe and secure, responsible and accountable and transparent and explainable
The characteristics outlined by the NIST AI RMF align well with Deloitte’s Trustworthy AI Framework and the focus on trustworthiness can help organizations to effectively utilize a variety of automated systems while feeling confident in the security and performance of their AI models.
Endnotes:
- National Institute of Standards and Technology AI Risk Management Framework
- The AI Bill of Rights follows the Executive Order 13960: Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government (December 2020)
- Executive Order 13859: Maintaining American Leadership in Artificial Intelligence (February 2019)
- Office of Management and Budget (OMB) Memorandum M-21-06: Guidance for Regulation of Artificial Intelligence Applications (November 2020), White House Office of Science and Technology Policy (OSTP): American AI Initiative: Year One Annual Report (February 2020)
- International initiatives include the Organisation for Economic Co-operation and Development (OECD): 2019 Recommendations on Artificial Intelligence, and the European Union Artificial Intelligence Act proposal (April 2021).
Get in touch
Oz Karan
Risk & Financial Advisory Trustworthy AI Leader
Partner
Deloitte & Touche LLP
Beena Ammanath
US Trustworthy and Ethical AI Leader
Global Deloitte AI Institute
Deloitte Consulting LLP
Ed Bowen
Advisory AI CoE Leader
Managing Director
Deloitte Consulting LLP
AI Center of Excellence
Let's talk
Recommendations
Trustworthy AI™
Bridging the ethics gap surrounding AI
Deloitte AI Institute
Connecting enterprises to all dimensions of the AI ecosystem