U.S. Advancing Responsible AI

The 21st century is frequently called the age of Artificial Intelligence (AI), prompting questions about its societal implications. It actively transforms numerous processes across various domains, and research ethics (RE) is no exception. Multiple challenges, encompassing accountability, privacy, and openness, are emerging.

Research Ethics Boards (REBs) have been instituted to guarantee adherence to ethical standards throughout research. This scoping review seeks to illuminate the challenges posed by AI in research ethics and assess the preparedness of REBs in evaluating these challenges. Ethical guidelines and standards for AI development and deployment are essential to address these concerns.

To sustain this awareness, the Oak Ridge National Laboratory (ORNL), a part of the Department of Energy, has joined the Trillion Parameter Consortium (TPC), a global collaboration of scientists, researchers, and industry professionals. The consortium aimed to address the challenges of building large-scale artificial intelligence (AI) systems and advancing trustworthy and reliable AI for scientific discovery.

ORNL’s collaboration with TPC aligns seamlessly with its commitment to developing secure, reliable, and energy-efficient AI, complementing the consortium’s emphasis on responsible AI. With over 300 researchers utilising AI to address Department of Energy challenges and hosting the world’s most powerful supercomputer, Frontier, ORNL is well-equipped to significantly contribute to the consortium’s objectives.

Leveraging its AI research and extensive resources, the laboratory will be crucial in addressing challenges such as constructing large-scale generative AI models for scientific and engineering problems. Specific tasks include creating scalable model architectures, implementing effective training strategies, organising and curating data for model training, optimising AI libraries for exascale computing platforms, and evaluating progress in scientific task learning, reliability, and trust.

TPC strives to build an open community of researchers developing advanced large-scale generative AI models for scientific and engineering progress. The consortium plans to voluntarily initiate, manage, and coordinate projects to prevent redundancy and enhance impact. Additionally, TPC seeks to establish a global network of resources and expertise to support the next generation of AI, uniting researchers focused on large-scale AI applications in science and engineering.

Prasanna Balaprakash, ORNL R&D staff scientist and director of the lab’s AI Initiative, said, “ORNL envisions being a critical resource for the consortium and is committed to ensuring the future of AI across the scientific spectrum.”

Further, as an international organisation that supports education, science, and culture, The United Nations Educational, Scientific and Cultural Organisation (UNESCO) has established ten principles of AI ethics regarding scientific research.

1. Beneficence: AI systems should be designed to promote the well-being of individuals, communities, and the environment.
2. Non-maleficence: AI systems should avoid causing harm to individuals, communities, and the environment.
3. Autonomy: Individuals should have the right to control their data and to make their own decisions about how AI systems are used.
4. Justice: AI systems should be designed to be fair, equitable, and inclusive.
5. Transparency: AI systems’ design, operation, and outcomes should be transparent and explainable.
6. Accountability: There should be clear lines of responsibility for developing, deploying, and using AI systems.
7. Privacy: The privacy of individuals should be protected when data is collected, processed, and used by AI systems.
8. Data security: Data used by AI systems should be secure and protected from unauthorised access, use, disclosure, disruption, modification, or destruction.
9. Human oversight: AI systems should be subject to human management and control.
10. Social and environmental compatibility: AI systems should be designed to be compatible with social and ecological values.

Since 1979, ORNL’s AI research has gained a portfolio with the launch of the Oak Ridge Applied Artificial Intelligence Project to ensure the alignment of UNESCO principles. Today, the AI Initiative focuses on developing secure, trustworthy, and energy-efficient AI across various applications, showcasing the laboratory’s commitment to advancing AI in fields ranging from biology to national security. The collaboration with TPC reinforces ORNL’s dedication to driving breakthroughs in large-scale scientific AI, aligning with the world agenda in implementing AI ethics.