As artificial intelligence (AI) and autonomous systems begin to pervade daily life, developers of these technologies must keep certain ethical considerations at hand to ensure the safety of human society.
IEEE announced three new standards for ethics in AI that prioritize human well-being as these technologies advance, according to a press release. They will become a part of the IEEE publication Ethically Aligned Design: "A Vision for Prioritizing Human Wellbeing with Artificial Intelligence and Autonomous Systems", a living document that encourages technologists to prioritize ethical considerations in working with AI.
The three IEEE standards projects are chaired by subject matter experts in their respective fields of study and they include the following:
1 - Standard for ethically driven nudging for robotic, intelligent, and autonomous systems
This standard examines "nudges," which in terms of AI are overt or hidden suggestions designed to influence human behavior or emotions. It explains the concepts, functions, and benefits needed to ensure that robotic and autonomous systems are adhering to worldwide ethics and moral theories. It emphasizes the need for aligning the ethics and engineering communities on designing and implementing these systems.
2 - Standard for fail-safe design of autonomous and semi-autonomous systems
Autonomous and semi-autonomous systems that malfunction can potentially harm human users, society, and the environment, IEEE noted. Effective fail-safe measures are needed to lower risks related to systems breaking down, and to provide developers, installers, and operators with clear technical instructions to terminate compromised systems safely. This standard establishes clear procedures for measuring, testing, and certifying an autonomous system's ability to fail safely on a scale from weak to strong, along with instructions for improving performance. It also provides a basis for developers, users, and regulators to design these fail-safe systems to improve accountability.
3 - Well-being metrics standard for ethical artificial intelligence and autonomous systems
As AI systems improve, programmers, engineers, and technologists must consider how the products and services they build can improve human well-being in terms of economic growth and productivity. This standard identifies human well-being indicators and metrics that may be directly impacted by autonomous and intelligent systems, and provides a baseline to align the data that these systems should include so they can be used to increase human well-being.