On May 1st, 2019, the National Institute of Standards and Technology (NIST) issued a Request for Information regarding the development of technical standards for artificial intelligence (AI) and the role of the federal government in this process. The Center for Democracy and Technology (CDT) submitted a comment in response, outlining important factors an AI standards-setting process should consider and corresponding recommendations.
In particular, we call for greater uniformity in the handling and securing of data, along with more standardized and transparent documentation of the datasets used in training AI and machine learning systems. Furthermore, we believe that a general framework and uniform vocabulary should be developed to facilitate the understanding and comparison of AI systems. Doing so will assist in policy creation and regulation, increase transparency, and promote the trustworthiness of AI. We also express concern about the potential for AI systems to automate bias and exacerbate inequality. To address this, we advocate for regular auditing of these systems to ensure that they are accurate, fair, and in compliance with legal and ethical standards.
CDT is excited about potential future collaboration with NIST, and believes that standardization will result in AI technologies that are powerful and innovative, while still serving the public good.