President Biden Issues Executive Order to Establish Standards for AI Technologies

 

By Lilian H. Hill

On October 30, 2023, President Biden issued an executive order to establish Artificial Intelligence (AI) safety and security standards, protect Americans’ privacy and security, advance equity and civil rights, and advocate for consumers and workers. It employs broad emergency powers, usually only invoked for urgent situations such as the coronavirus pandemic or war, and the power of multiple government agencies to address the risks of artificial intelligence, which Biden described as the “most consequential technology of our time.” Biden has also called on Congress to create legislation to regulate AI as multiple attempts have failed to pass.

 

Released just days before an international AI Safety Summit held in the UK (Zakrzewski et al., 2023), the 111-page Executive Order has seven focus areas: 

• safety
• protection of American’s privacy 
• preventing bias 
• supporting consumers, students, and patients 
• supporting workers 
• promoting innovation 
• advancing American leadership abroad

 

Selected details are described for each focus area below.

 

Safety

 

• AI corporations will be required to conduct safety assessments of their products and submit findings to the federal government before implementing AI technology.
• Safeguards should be in place to shield Americans from AI-facilitated fraud and deception.
• Protocols and best practices will be established to detect AI-generated content and validate official content.
• A sophisticated cybersecurity initiative will be developed to identify and rectify vulnerabilities in critical software.

Protection of Americans Privacy

• Federal backing for expediting the development and application of privacy-preserving methods will be utilized, incorporating the use of cryptographic tools.
• Efforts will be made to strengthen the methods by which federal agencies collect and use commercially available information, alongside privacy guidelines to tackle AI-related risks.

Preventing Bias

• Standards will be formulated to furnish landlords, federal benefits programs, and federal contractors with precise guidelines to prevent AI algorithms from exacerbating discrimination.
• The Department of Justice and federal civil rights personnel will receive training to address algorithmic discrimination and to adopt best practices for investigating and prosecuting AI-related civil rights violations.
• Fairness will be encouraged throughout the criminal justice system by outlining best practices for the use of AI in sentencing, parole and probation, pretrial release and detention, risk assessments, surveillance, crime forecasting and predictive policing, and forensic analysis.

Stand Up for Consumers, Students, and Patients

• The responsible use of AI in healthcare will be promoted, with the Department of Health and Human Services establishing a safety program to receive reports of, and take action against, harms or unsafe healthcare practices involving AI.
• Efforts will be made to shape AI's potential to revolutionize education by creating resources to support educators deploying AI-driven educational tools, such as personalized tutoring in schools.

Promoting Innovation

• AI research across the United States will receive support by initiating the National AI Research Resource pilot program, providing AI researchers and students access to vital AI resources and data and increased grants for AI research in critical areas such as healthcare and climate change.
• An equitable, open, and competitive AI ecosystem will be encouraged by granting small developers and entrepreneurs access to technical aid and resources, assisting small businesses in commercializing AI breakthroughs, and encouraging the Federal Trade Commission to exercise its authority.
• The opportunities for highly skilled immigrants and nonimmigrants with expertise in crucial fields to study, remain, and work in the United States will be expanded by modernizing and streamlining visa criteria, interviews, and reviews.

Supporting Workers

• Principles and best practices will be devised to mitigate the negative impacts and maximize the benefits of AI for workers, addressing job displacement, labor standards, workplace equity, health and safety, and data collection.
• A report will be compiled on AI's potential effects on the labor market, and strategies to bolster federal support for workers facing labor disruptions, including those resulting from AI, will be developed.
• The opportunities for highly skilled immigrants and nonimmigrants with expertise in crucial fields to study, remain, and work in the United States will be expanded by modernizing and streamlining visa criteria, interviews, and reviews.

Advancing American Leadership Abroad

• Strengthen bilateral, multilateral, and multi-stakeholder engagements to collaborate on AI with the State Department, in conjunction with the Commerce Department, leading efforts to establish robust international frameworks for harnessing AI's benefits and managing its risks, ensuring safety.
• Accelerate the development and implementation of crucial AI standards with international partners and in standards organizations, ensuring the technology's safety, security, trustworthiness, and interoperability.
• Advocate for the safe, responsible, and rights-affirming development and deployment of AI globally to tackle global challenges, such as advancing sustainable development and mitigating threats to critical infrastructure.

Ensuring Responsible and Effective Government Use of AI

• Issue guidelines for agencies' use of AI, including clear standards to protect rights and safety, improve AI procurement, and strengthen AI deployment.
• Assist agencies in procuring specified AI products and services more quickly, affordably, and effectively through streamlined contracting processes.
• Expedite the rapid hiring of AI professionals as part of a government-wide AI talent surge led by the Office of Personnel Management, U.S. Digital Service, U.S. Digital Corps, and Presidential Innovation Fellowship. Agencies will provide AI training for employees at all levels in relevant fields.

 

This Executive Order is far-reaching and comprehensive. It builds on and expands the considerations articulated in the White House Blueprint for an AI Bill of Rights. However, news commentary indicates that the Executive Order is only a first step in regulating rapid AI development and implementation, but also indicates that the Executive Order needs to be improved in enforcement mechanisms.

 

References

The White House. (2023, October 30). Fact Sheet: President Biden Issues Executive Order on Safe, Secure, and Trustworthy Artificial Intelligence. Retrieved from https://www.whitehouse.gov/briefing-room/statements-releases/2023/10/30/fact-sheet-president-biden-issues-executive-order-on-safe-secure-and-trustworthy-artificial-intelligence/

The White House Office of Science and Technology Policy (n.d.). Blueprint for an AI Bill of Rights: Making Automated Systems Work for the American People. Retrieved https://www.whitehouse.gov/ostp/ai-bill-of-rights/#safe

Zakrzewski, C., Lima, C., & Pager, T. (2023, October 25). White House to unveil sweeping AI executive order next week. Washington Post. Retrieved https://www.washingtonpost.com/technology/2023/10/25/artificial-intelligence-executive-order-biden/?itid=ap_catzakrzewski

Efforts to Regulate Algorithms

 

We have discussed the dangers embedded in algorithms in previous episodes. Therefore, there is a need for oversight and regulation. On this episode, Dr. Lilian Hill talks about the efforts to regulate algorithms, provides a historical perspective, and explains what is being done in the U.S. and other countries.

Listen to the Podcast

Information Literacy Episode 7 Transcript

 

References

Ferguson, C., P’ng, J., & Whiteside, H. (2022, October 18). The Regulation of Artificial Intelligence in Canada and Abroad: Comparing the Proposed AIDA and EU AI Act. Fasken. Retrieved https://www.fasken.com/en/knowledge/2022/10/18-the-regulation-of-artificial-intelligence-in-canada-and-abroad

Government of Canada. Algorithmic Impact Assessment tool. Department of Treasury. Retrieved https://www.canada.ca/en/government/system/digital-government/digital-government-innovations/responsible-use-ai/algorithmic-impact-assessment.html

Newton, C. (2020, December 29). Everything you need to know about Section 230: The most important law for online speech, The Verge. Retrieved https://www.theverge.com/21273768/section-230-explained-internet-speech-law-definition-guide-free-moderation 

White House Office of Science and Technology Policy (n.d.). Blueprint for an AI Bill Of rights: Making Automated Systems Work for the American People. Retrieved https://www.whitehouse.gov/ostp/ai-bill-of-rights/#safe

 

 

 

Dangers and Recommendations of Algorithms

 

By Lilian H. Hill

A lot of uses of algorithms are beneficial; however, there are dangers involved. Decisions about admissions, scholarship awards, and hiring have been turned over to algorithms. Flaws in programming can cost individuals the opportunity to attend their college of choice because their entrance exams were graded based on a faulty metric. Hiring decisions can be negatively influenced if metrics involved are biased against minorities. People’s privacy can be violated if the algorithms designed to share individual’s information are inaccurate. 

In an interesting news story, Northeastern University installed heat sensor devices undergraduate students’ desks to track usage (Ongweso, 2022). Given that the students were enrolled in Northeaster’s Cybersecurity and Privacy Institute, it should not be surprising that the students detected the presence of the devices, hacked into them, developed an open-source guide so that other students could hack them. They then removed the devices and displayed them in an art exhibit spelling the word NO! The university had installed the devices at night without informing the students and without Institutional Review Board (IRB) permission. The students found that the devices were not as secure as the university claimed. 

Recommendations

The Center for Democracy and Technology recommends the following:

• Human beings need to retain control of decision-making that involves people’s privacy, safety, and opportunities. Context and nuance are difficult to program into algorithms. 
• Regulate data governance: Establish policies that determine long information should be kept and under what certain conditions it should be deleted.
• Conduct regular audits to ensure that discriminatory outcomes or other unexpected harm do not occur.
• Communicate regularly with stakeholders to provide feedback and address concerns about the systems that affect their schools.
• Use algorithms for the purposes they were designed for. Adapting them to other purposes has the potential to yield harmful results.
• Foster accountability by developing plans and policies to identify and correct errors in the programming. Have strategies and resources available to make amends when errors have been harmful to people. 
• Ensure legal compliance so that the decisions made by algorithms are fair, accurate, and comply with legal standards for education. 

References

Center for Democracy and Technology. Algorithmic systems in education: Incorporating equity and fairness when using student data. Retrieved from https://cdt.org/insights/algorithmic-systems-in-education-incorporating-equity-and-fairness-when-using-student-data/ 

Ongweso, E. (2022, December 2). ‘NO’: Grad students analyze, hack, and remove under-desk surveillance devices designed to track them. https://www.vice.com/en/article/m7gwy3/no-grad-students-analyze-hack-and-remove-under-desk-surveillance-devices-designed-to-track-them