![\"\"](\"https:\/\/world.edu\/wp-content\/uploads\/2023\/11\/BidenExecutiveOrder.jpg\")
<\/p>\n<\/p>\n
The\u00a0comprehensive, even sweeping, set of guidelines<\/a>\u00a0for artificial intelligence that the White House unveiled in an executive order on Oct. 30, 2023, show that the U.S. government is attempting to address the risks posed by AI.<\/p>\n As a\u00a0researcher of information systems and responsible AI<\/a>, I believe the executive order represents an important step in building\u00a0responsible<\/a>\u00a0and\u00a0trustworthy<\/a>\u00a0AI.<\/p>\n The order is only a step, however, and it leaves unresolved the issue of comprehensive data privacy legislation. Without such laws, people are at greater risk of\u00a0AI systems revealing sensitive or confidential information<\/a>.<\/p>\n Technology is typically evaluated for\u00a0performance, cost and quality<\/a>, but often not equity, fairness and transparency. In response, researchers and practitioners of responsible AI have been advocating for:<\/p>\n The National Institute of Standards and Technology (NIST) issued a\u00a0comprehensive AI risk management framework<\/a>\u00a0in January 2023 that aims to address many of these issues. The framework\u00a0serves as the foundation<\/a>\u00a0for much of the Biden administration\u2019s executive order. The executive order also\u00a0empowers the Department of Commerce<\/a>, NIST\u2019s home in the federal government, to play a key role in implementing the proposed directives.<\/p>\n Researchers of AI ethics have long cautioned that\u00a0stronger auditing of AI systems<\/a>\u00a0is needed to avoid giving the appearance of scrutiny\u00a0without genuine accountability<\/a>. As it stands, a recent study looking at public disclosures from companies found that claims of AI ethics practices\u00a0outpace actual AI ethics initiatives<\/a>. The executive order could help by specifying avenues for enforcing accountability.<\/p>\n Another important initiative outlined in the executive order is probing for vulnerabilities of\u00a0very large-scale general-purpose AI models<\/a>\u00a0trained on massive amounts of data, such as the models that power OpenAI\u2019s ChatGPT or DALL-E. The order requires companies that build large AI systems with the potential to affect national security, public health or the economy\u00a0to perform red teaming<\/a>\u00a0and report the results to the government. Red teaming is using manual or automated methods to attempt to\u00a0force an AI model to produce harmful output<\/a>\u00a0\u2013 for example, make offensive or dangerous statements like advice on how to sell drugs.<\/p>\n Reporting to the government is important given that a recent study found\u00a0most of the companies that make these large-scale AI systems lacking<\/a>\u00a0when it comes to transparency.<\/p>\n Similarly, the public is at risk of being fooled by AI-generated content. To address this, the executive order directs the Department of Commerce to\u00a0develop guidance for labeling AI-generated content<\/a>. Federal agencies will be required to use\u00a0AI watermarking<\/a>\u00a0\u2013 technology that marks content as AI-generated to reduce fraud and misinformation \u2013 though it\u2019s not required for the private sector.<\/p>\n The executive order also\u00a0recognizes that AI systems can pose unacceptable risks<\/a>\u00a0of\u00a0harm to civil and human rights<\/a>\u00a0and the well-being of individuals: \u201cArtificial Intelligence systems deployed irresponsibly have reproduced and intensified existing inequities, caused new types of harmful discrimination, and exacerbated online and physical harms.\u201d<\/p>\nUnderstanding AI risks<\/h2>\n
\n