2024-12

Supporting Responsible Development and Deployment of Artificial Intelligence

Sponsored by

Rep. Christopher Rosario (CT)

 

Reported to the Caucus by the NHCSL Broadband and Technology Task Force

Rep. Arturo Alonso Sandoval (OK), Chair

Unanimously ratified by the Caucus on November 23, 2024

WHEREAS, artificial intelligence (AI) has the potential to bring about significant benefits to society, but its development and deployment also raise important ethical and social implications; and,

WHEREAS, algorithmic or training-data[1] discrimination and bias in AI systems can perpetuate and exacerbate existing inequalities, even when unintended;[2] and,

WHEREAS, ethical AI is “the practice of designing and using AI systems to align with ethical principles and human values, considering their potential impact on society;”[3] and,

WHEREAS, algorithmic fairness is a set of “efforts to ensure AI systems make decisions without discriminating against certain groups or individuals;”[4] and,

WHEREAS, ensuring that AI systems are designed and used in ways that promote fairness, transparency, and accountability is crucial for building trust and promoting social good.

THEREFORE, BE IT RESOLVED, the National Hispanic Caucus of State Legislators recognizes the need for responsible development and deployment of artificial intelligence (AI) that promotes fairness, transparency and accountability, and supports legislation that promotes or feasibly seeks to achieve Ethical AI and Algorithmic Fairness, prevent algorithmic or training-data bias and discrimination and ensure accountability, or legislation that incentivizes the development of AI systems that benefit society as a whole and promote the well-being of all individuals; and,

BE IT FURTHER RESOLVED, that the National Hispanic Caucus of State Legislators supports initiatives that promote diversity, equity, and inclusion in the development and deployment of AI, and that address the potential impacts of AI on marginalized communities; and,

BE IT FURTHER RESOLVED, that the National Hispanic Caucus of State Legislators encourages developers and deployers of AI systems to take steps to mitigate the risks of algorithmic and training-data bias and discrimination and to ensure that their systems are used in ways that promote fairness, transparency, explainability and accountability; and,

BE IT FURTHER RESOLVED, NHCSL supports the following provisions:

• Definitions of key terms[5] in a standard manner to ensure ease of accountability and compliance;
• Requirements for developers to use reasonable care to protect consumers from known or reasonably-foreseeable risks of algorithmic discrimination;
• Requirements for developers to provide documentation and information to deployers about the intended uses, limitations, and potential biases of high-risk artificial intelligence systems;
• Requirements for developers to provide deployers with the information necessary to complete impact assessments on the potential risks and benefits of high-risk artificial intelligence systems; and,

BE IT FURTHER RESOLVED, that the National Hispanic Caucus of State Legislators shall continue monitoring AI developments and identify potential impacts on Hispanic communities and collaborate with other organizations and stakeholders to advance responsible AI development and deployment.

THE NHCSL BROADBAND AND TECHNOLOGY TASK FORCE, AT ITS MEETING OF JULY 9, 2024, UNANIMOUSLY RECOMMENDED THIS RESOLUTION TO THE EXECUTIVE COMMITTEE FOR APPROVAL.

THE EXECUTIVE COMMITTEE UNANIMOUSLY APPROVED THIS RESOLUTION AT ITS MEETING OF JULY 13, 2024.

THE NATIONAL HISPANIC CAUCUS OF STATE LEGISLATORS UNANIMOUSLY AMENDED AND RATIFIED THIS RESOLUTION AT ITS ANNUAL MEETING OF NOVEMBER 23, 2024 IN DENVER, COLORADO.

[1] AI and You, AI Basics Glossary (see Training Data)

[2] AI and You, AI Basics Glossary (see Bias)

[3] AI and You, AI Basics Glossary

[4] Ibid.

[5] Examples include "algorithmic discrimination", "artificial intelligence system", "consequential decision", "consumer", "deploy", "deployer", "developer", "general-purpose artificial intelligence model", "generative artificial intelligence system", "high-risk artificial intelligence system", "intentional and substantial modification", "person", and "synthetic digital content"