On Monday, President Biden made public a long-awaited comprehensive executive order on “artificial intelligence” (AI), with the aim of capitalizing on this emerging technology while also mitigating its potential risks.
The executive order outlines a series of new initiatives that focus on areas such as “safety,” “privacy,” “protecting workers,” and “protecting innovation.”
A high-ranking official in the administration communicated to the media, “President Biden is committed to leveraging the benefits of AI for the public good, while also safeguarding individuals from its potentially significant risks.”
The executive order sets forth new “safety” standards, including the requirement for companies developing models that could pose a substantial risk to “national security,” “economic security,” or “public health” to inform the federal government during the model’s training phase and to disclose the results of all “safety tests.” The Commerce Department is also assigned the task of creating guidelines for “content authentication” and “watermarking” to label AI-generated content.
The executive order instructs federal agencies to speed up the development of methods that allow AI systems to be trained while preserving the “privacy” of the training data. It also aims to scrutinize how these agencies gather and utilize commercially available information containing “personal data.”
To back “workers,” the executive order establishes principles and best practices to minimize the negative impacts and maximize the benefits of AI for employees, addressing issues like “job displacement,” “labor standards,” and “data collection.”
The executive order also aims to stimulate “innovation” and competition by initiating a pilot of a tool, the National AI Research Resource, which will increase grants for AI research in sectors such as “health care” and “climate change.” It will also utilize existing laws to expand the opportunities for highly skilled immigrants and non-immigrants with expertise in these critical sectors to study, reside, and work in the United States by simplifying “visa criteria” and interviews.
The executive order is applicable to companies with the most potent AI systems, regardless of their affiliations with the federal government, an official stated. This marks the most substantial effort to establish a national framework for the AI sector, especially after the advent of ChatGPT, which astonished the public with its human-like capabilities less than a year ago.
The executive order is designed to make the U.S. a leader in AI and to enhance American global leadership in this emerging technology. Vice President Harris is set to travel to the United Kingdom on Tuesday to discuss the U.S. vision for AI.
Administration officials, however, reject the notion that the U.S. is trailing Europe in AI regulation. “We’re not in a race, nor are we lagging,” said the official. “I would counter any suggestion that we’re behind anyone. Ultimately, this is an opportunity for collaborative governance in AI. I don’t see a significant amount of tension between the United States and Europe on this issue.”
The executive order focuses on advancing “equity” and “civil rights” by offering guidance to landlords, federal benefits programs, and federal contractors to prevent AI algorithms from exacerbating “discrimination.” It also aims to create a set of best practices for the use of AI in the “criminal justice system,” covering areas like “sentencing,” “parole and probation,” “surveillance,” “crime forecasting,” and “forensic analysis.”
Furthermore, the executive order instructs the Department of Health and Human Services to establish a “safety program” to receive reports of unsafe healthcare practices involving AI and to create resources to aid educators in deploying AI tools for educational purposes.
“President Biden is initiating the most robust set of actions any government in the world has ever taken concerning AI safety, security, and trust,” Bruce Reed, the deputy chief of staff, stated.
To achieve the goals of the executive order, the White House will form an AI council for principal-level meetings, spearheaded by Reed. The timeline for the implementation of the executive order varies, with some “safety actions” to be enacted within 90 days and other tasks to be completed within 270 days or a year.
Additionally, the White House is launching AI.gov for individuals interested in AI-related careers in the federal government. The official noted that they aim to bring more AI expertise into the federal workforce, stating, “you can never have too much AI talent.”