Artificial Intelligence: Backgrounds, Risks, and Policies

Contents
Preface
Chapter 1
Artificial Intelligence: Background, Selected Issues, and Policy Considerations(
	Summary
	Introduction
	What Is AI?
		AI Terminology
		Algorithms and AI
	Historical Context of AI
		Waves of AI
	Recent Growth in the Field of AI
		AI Research and Development
		Private and Public Funding
	Selected Research and Focus Areas
		Explainable AI
		Data Access
		AI Training with Small and Alternative Datasets
		AI Hardware
	Federal Activity in AI
		Executive Branch
			Executive Orders on AI
			National Science and Technology Council Committees
			Select AI Reports and Documents
			Federal Agency Activities
		Congress
			Legislation
			Hearings
	Selected Issues for Congressional Consideration
		Implications for the U.S. Workforce
			Job Displacement and Skill Shifts
			AI Expert Workforce
		International Competition and Federal Investment in AI R&D
		Standards Development
		Ethics, Bias, Fairness, and Transparency
			Types of Bias
Chapter 2
Trustworthy AI: Managing the Risks of Artificial Intelligence *
	U.S. House of Representatives, Committee on Science, Space, and Technology, Subcommittee on Research and Technology, Hearing Charter, Trustworthy AI: Managing the Risks  of Artificial Intelligence
		Purpose
		Witnesses
		Overarching Questions
		Background
		AI Risks
			Harmful Bias
			Explainability and Interpretability
			Safety
			Cybersecurity and Privacy
			Computational Costs
		Government Action
			OSTP
			National Institute of Standards and Technology
			National Science Foundation
			International
		Private Sector Action
	Testimony of Ms. Elham Tabassi, Chief of staff, Information Technology Laboratory, National Institute of Standards  and Technology
	Testimony of Elham Tabassi, Chief of Staff, Information Technology Laboratory, National Institute of Standards and Technology, United States Department of Commerce, before the United States House of Representatives, Committee on Science, Space, and Te...
		NIST’s Role in Artificial Intelligence
		NIST AI Risk Management Framework
		NIST’s Research on AI Trustworthiness Characteristics
			AI Trustworthiness Characteristics – Fair and Bias is Managed
			AI Trustworthiness Characteristics – Explainable and Interpretable
			AI Trustworthiness Characteristics –Secure and Resilient
			AI Trustworthiness Characteristics – Privacy-enhanced
		Research on Applications of AI
		AI Measurement and Evaluation
		AI Standards
		Interagency Coordination
		Conclusion
	Elham Tabassi (Fed), Chief of Staff, Information  Technology Laboratory
	Testimony of Dr. Charles Isbell, Dean and John P. Imlay,  Jr. Chair of the College of Computing, Georgia Institute  of Technology
	Testimony of Mr. Jordan Crenshaw, Vice President of the Chamber Technology Engagement Center, U.S. Chamber  of Commerce
	Before the U.S. House Research And Technology Subcommittee, Hearing on “Trustworthy AI:  Managing the Risks of Artificial Intelligence,”  Testimony of Jordan Crenshaw, Vice President,  C_TEC, U.S. Chamber of Commerce, September 29, 2022
		Opportunities for the Federal Government and Industry to Work Together to Develop Trustworthy AI
			Congress Needs to Pass a Preemptive National Data Privacy Law
			Support for Alternative Regulatory Pathways Such as Voluntary Consensus Standards
			Stakeholder Driven Engagement
			Awareness of the Benefits of Artificial Intelligence
			Awareness of the Benefits of Artificial Intelligence
			How Are Different Sectors Adopting Governance Models and Other Strategies to Mitigate Risks that Arise from AI Systems?
			How Should the United States Encourage More Organizations to Think Critically about Risks that Arise from AI Systems, Including by Priortiziing Trustworthy AI from the Earliest Stages of Development of New Systems?
			What Recommendations Do You Have for how the Federal Government can Strengthen its Role for the Development and Responsible Deployment of Trustworthy AI Systems?
		Conclusion
	Testimony of Ms. Navrina Singh, Founder and Chief Executive Officer, Credo AI
	Prepared Testimony of Navrina Singh, Founder and CEO, Credo AI, before the House Committee on Science, Space and Technology, Subcommittee on Research and Technology
		Introduction
		What Is Responsible AI?
		How to Create an Environment that Fosters RAI
		Companies Are Seeking Guidance
		Key Challenges to Overcome in the Development and Use of Responsible AI
		Context Is Critical: Metrics for Each Tenant of RAI Vary
		Addressing Risk Now Ensures Leadership in the Long Run
	Conclusion
	Appendix I: Answers to Post-Hearing Questions
	Appendix II: Additional Material for the Record
		Engineered Intelligence: Creating a Successor Species,  Congressman Brad Sherman, Statement for  the Committee on Science, Space, & Technology,  May 17, 2019
Chapter 3
Blueprint for an AI Bill of Rights:  Making Automated Systems Work for  the American People, October 2022*
	Foreword
	About This Framework
	Listening to the American Public
	Blueprint for an AI Bill of Rights
		Safe and Effective Systems
			You Should Be Protected from Unsafe or Ineffective Systems
		Algorithmic Discrimination Protections
			You Should Not Face Discrimination by Algorithms and Systems Should Be Used and Designed in an Equitable Way
		Data Privacy
			You Should Be Protected from Abusive Data Practices via Built-In Protections and You Should Have Agency over How Data About You  Is Used
		Notice and Explanation
			You Should Know That an Automated System Is Being Used and Understand How and Why It Contributes to Outcomes That  Impact You
		Human Alternatives, Consideration, and Fallback
			You Should Be Able to Opt out, Where Appropriate, and Have Access to a Person Who Can Quickly Consider and Remedy Problems  You Encounter
	Applying the Blueprint for an AI Bill of Rights
		Rights, Opportunities, or Access
		Relationship to Existing Law and Policy
	Applying the Blueprint for an AI Bill of Rights
		Relationship to Existing Law and Policy
		Definitions
			Algorithmic Discrimination
			Automated System
			Communities
			Equity
			Rights, Opportunities, or Access
			Sensitive Data
			Sensitive Domains
			Surveillance Technology
			Underserved Communities
	From Principles to Practice: A Techincal Companion to the Blueprint for an AI Bill of Rights
		Using This Technical Companion
	Safe and Effective Systems
		You Should Be Protected from Unsafe or Ineffective Systems
		Why This Principle Is Important
		What Should Be Expected of Automated Systems
			Protect the Public from Harm in a Proactive and Ongoing Manner
				Consultation
				Testing
				Risk Identification and Mitigation
				Ongoing Monitoring
				Clear Organizational Oversight
			Avoid Inappropriate, Low-Quality, or Irrelevant Data Use and the Compounded Harm of Its Reuse
				Relevant and High-Quality Data
				Derived Data Sources Tracked and Reviewed Carefully
				Data Reuse Limits in Sensitive Domains
			Demonstrate the Safety and Effectiveness of the System
				Independent Evaluation
				Reporting
		How These Principles Can Move into Practice
			Executive Order 13960 on Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government Requires That Certain Federal Agencies Adhere to Nine Principles When Designing, Developing, Acquiring, or Using AI for Purposes Other Than Nat...
			The Law and Policy Landscape for Motor Vehicles Shows That  Strong Safety Regulations—and Measures to Address Harms  When They Occur—Can Enhance Innovation in the Context of Complex Technologies
			From Large Companies to Start-Ups, Industry Is Providing Innovative Solutions That Allow Organizations to Mitigate Risks to the Safety and Efficacy of AI Systems, Both before Deployment and through Monitoring over Time
			The Office of Management and Budget (OMB) Has Called for an Expansion of Opportunities for Meaningful Stakeholder Engagement in the Design of Programs and Services
			The National Institute of Standards and Technology (NIST) Is Developing a Risk Management Framework to Better Manage Risks Posed to Individuals, Organizations, and Society by AI
			Some U.S Government Agencies Have Developed Specific Frameworks for Ethical Use of AI Systems
			The National Science Foundation (NSF) Funds Extensive Research to Help Foster the Development of Automated Systems That Adhere to and Advance Their Safety, Security and Effectiveness
			Some State Legislatures Have Placed Strong Transparency and Validity Requirements on the Use of Pretrial Risk Assessments
	Algorithmic Discrimination Protections
		You Should Not Face Discrimination by Algorithms and Systems Should Be Used and Designed in an Equitable Way
		Why This Principle Is Important
		What Should Be Expected of Automated Systems
			Protect the Public from Algorithmic Discrimination in a Proactive and Ongoing Manner
				Proactive Assessment of Equity in Design
				Representative and Robust Data
				Guarding against Proxies
				Ensuring Accessibility during Design, Development, and Deployment
				Disparity Assessment
				Disparity Mitigation
				Ongoing Monitoring and Mitigation
			Demonstrate That the System Protects against Algorithmic Discrimination
				Independent Evaluation
				Reporting
		How These Principles Can Move into Practice
			The Federal Government Is Working to Combat Discrimination in Mortgage Lending
			The Equal Employment Opportunity Commission and the Department of Justice Have Clearly Laid out How Employers’ Use of AI and Other Automated Systems Can Result in Discrimination against Job Applicants and Employees with disabilities
			Disparity Assessments Identified Harms to Black Patients\'  Healthcare Access
			Large Employers Have Developed Best Practices to Scrutinize the Data and Models Used for Hiring
			Standards Organizations Have Developed Guidelines to Incorporate Accessibility Criteria into Technology Design Processes
			NIST Has Released Special Publication 1270, towards a Standard for Identifying and Managing Bias in Artificial Intelligence
	Data Privacy
		You Should Be Protected from Abusive Data Practices via Built-in Protections and You Should Have Agency over How Data About You Is Used
		Why This Principle Is Important
		What Should Be Expected of Automated Systems
			Protect Privacy by Design and by Default
				Privacy by Design and by Default
				Data Collection and Use-Case Scope Limits
				Risk Identification and Mitigation
				Privacy-Preserving Security
				Protect the Public from Unchecked Surveillance
					Heightened Oversight of Surveillance
					Limited and Proportionate Surveillance
					Scope Limits on Surveillance to Protect Rights and Democratic Values
				Provide the Public with Mechanisms for Appropriate and Meaningful Consent, Access, and Control over Their Data
					Use-Specific Consent
					Brief and Direct Consent Requests
					Data Access and Correction
					Consent Withdrawal and Data Deletion
					Automated System Support
				Demonstrate That Data Privacy and User Control Are Protected
					Independent Evaluation
					Reporting
		Extra Protections for Data Related to Sensitive Domains
		What Should Be Expected of Automated Systems
			Provide Enhanced Protections for Data Related to Sensitive Domains
				Necessary Functions Only
				Ethical Review and Use Prohibitions
				Data Quality
				Limit Access to Sensitive Data and Derived Data
				Reporting
		How These Principles Can Move into Practice
			The Privacy Act of 1974 Requires Privacy Protections for Personal Information in Federal Records Systems, Including Limits on Data Retention, and Also Provides Individuals a General Right to Access and Correct Their Data
			NIST’s Privacy Framework Provides a Comprehensive, Detailed and Actionable Approach for Organizations to Manage Privacy Risks
			A School Board’s Attempt to Surveil Public School Students—Undertaken without Adequate Community Input—Sparked a State-Wide Biometrics Moratorium
			Federal Law Requires Employers, and Any Consultants They May Retain, to Report the Costs of Surveilling Employees in the Context of a Labor Dispute, Providing a Transparency Mechanism to Help Protect Worker Organizing
			Privacy Choices on Smartphones Show That When Technologies Are Well Designed, Privacy and Data Agency Can Be Meaningful and  Not Overwhelming
	Notice and Explanation
		You Should Know That an Automated System Is Being Used,  and Understand How and Why It Contributes to Outcomes That Impact You
		Why This Principle Is Important
		What Should Be Expected of Automated Systems
			Provide Clear, Timely, Understandable, and Accessible Notice of Use and Explanations
				Generally Accessible Plain Language Documentation
				Accountable
				Timely and up-to-Date
				Brief and Clear
			Provide Explanations as to How and Why a Decision Was Made or an Action Was Taken by an Automated System
				Tailored to the Purpose
				Tailored to the Target of the Explanation
				Tailored to the Level of Risk
				Valid
			Demonstrate Protections for Notice and Explanation
				Reporting
		How These Principles Can Move into Practice
			Real-Life Examples of How These Principles Can Become Reality, Through Laws, Policies, and Practical Technical and Sociotechnical Approaches to Protecting Rights, Opportunities, and Access
				People in Illinois Are Given Written Notice by the Private Sector if Their Biometric Information Is Used
				Major Technology Companies Are Piloting New Ways to Communicate with the Public About Their Automated Technologies
				Lenders Are Required by Federal Law to Notify Consumers About Certain Decisions Made About Them
				A California Law Requires That Warehouse Employees Are Provided with Notice and Explanation About Quotas, Potentially Facilitated by Automated Systems, That Apply to Them
				Across the Federal Government, Agencies Are Conducting and Supporting Research on Explainable AI Systems
	Human Alternatives, Consideration, and Fallback
		You Should Be Able to Opt out, Where Appropriate, and Have Access to a Person Who Can Quickly Consider and Remedy Problems You Encounter
		Why This Principle Is Important
		What Should Be Expected of Automated Systems
			Provide a Mechanism to Conveniently Opt out from Automated Systems in Favor of a Human Alternative, Where Appropriate
				Brief, Clear, Accessible Notice and Instructions
				Human Alternatives Provided When Appropriate
				Timely and Not Burdensome Human Alternative
			Provide Timely Human Consideration and Remedy by a Fallback and Escalation System in the Event That an Automated System Fails, Produces Error, or You Would Like to Appeal or Contest Its Impacts on You
				Proportionate
				Accessible
				Convenient
				Equitable
				Timely
				Effective
				Maintained
			Institute Training, Assessment, and Oversight to Combat Automation Bias and Ensure any Human-Based Components of a System  Are Effective
				Training and Assessment
				Oversight
			Implement Additional Human Oversight and Safeguards for Automated Systems Related to Sensitive Domains
				Narrowly Scoped Data and Inferences
				Tailored to the Situation
				Human Consideration before Any High-Risk Decision
				Meaningful Access to Examine the System
			Demonstrate Access to Human Alternatives, Consideration,  and Fallback
				Reporting
		How These Principles Can Move into Practice
			Healthcare “Navigators” Help People Find Their Way through Online Signup Forms to Choose and Obtain Healthcare
			The Customer Service Industry Has Successfully Integrated Automated Services Such as Chat-Bots and AI-Driven Call Response Systems with Escalation to a Human Support Team
			Ballot Curing Laws in at Least 24 States Require a Fallback System That Allows Voters to Correct Their Ballot and Have It Counted in the Case That a Voter Signature Matching Algorithm Incorrectly Flags Their Ballot as Invalid or There Is Another Issue...
	Appendix
		Examples of Automated Systems
		Listening to the American People
		Panel Discussions to Inform the Blueprint for an AI Bill of Rights
		Summaries of Panel Discussions
			Panel 1: Consumer Rights and Protections
				Welcome
				Moderator
				Panelists
				Panel 2: The Criminal Justice System
					Welcome
					Moderator
					Panelists
				Panel 3: Equal Opportunities and Civil Justice
					Welcome
					Moderator
					Panelists
				Panel 4: Artificial Intelligence and Democratic Values
					Welcome
					Moderator
					Panelists
				Panel 5: Social Welfare and Development
					Welcome
					Moderator
					Panelists
				Panel 6: The Healthcare System
					Welcome
					Moderator
					Panelists
Index
Blank Page
Blank Page