IEEE Standards Association (IEEE-SA) released the first version of Ethics guidelines for automation and Intelligent systems, titled “Ethically Aligned Design (EAD): A vision for Prioritizing Human Well-being with Autonomous and Intelligent Systems”, earlier this week.
EAD guidelines feature scientific analysis and resources, high-level principles as well as actionable recommendations for ethical implementation of autonomous and intelligent systems (A/IS). “We offer high-level General Principles in Ethically Aligned Design that we consider to be imperatives for creating and operating A/IS that further human values and ensure trustworthiness”, reads EAD.
The EAD guideline explains eight high-level ethical principles that can be applied to all types of autonomous and intelligent systems (A/IS), irrespective of whether they are physical robots, software systems or algorithmic chatbots.
Eight General Principles in EAD
As mentioned in EAD, A/IS shall be created and operated in such a way that it respects, promotes, and protects internationally the recognized human rights. These rights should be fully taken into consideration by individuals, companies, research institutions, and governments to reflect the principle that A/IS respects and fulfills the human rights, freedoms, human dignity, and cultural diversity.
EAD states that A/IS creators should focus on improving human well-being as a primary success criterion for development. EAD recommends that A/IS should prioritize human well-being as the outcome in all system designs. It should use the best available and widely accepted “well-being metrics” as their reference point.
A/IS creators should put more emphasis on empowering individuals with an added ability to access and securely share their data. A/IS creators should focus on maintaining people’s capacity to have control over their identity.
Organizations and governments, should test and implement technologies that allow the individuals to specify their online agent for case-by-case authorization decisions. For minors, current guardianship approaches should be implemented to determine their suitability in this context.
Creators should provide evidence of the effectiveness and fitness for the purpose of A/IS. EAD recommends that creators engaged in the development of A/IS should focus on defining the metrics to serve as valid and meaningful gauges of the effectiveness of the system.
Creators of A/IS should design systems where the metrics on specific deployments of the system can be aggregated to deliver information on the effectiveness of the system across different deployments. Also, industry associations and other organizations (IEEE and ISO) should collaborate to develop standards for reporting on the effectiveness of A/IS.
EAD states that the basis of a particular A/IS decision should always be discoverable. It recommends that new standards should be developed in a way that it describes measurable and testable levels of transparency.
Also, these standards would offer designers with a guide for self-assessing transparency during development and suggest mechanisms for improving transparency.
As per EAD, A/IS should be created and operated in a way so that it offers an “unambiguous rationale” for decisions made. EAD states that in order to address the issues of responsibility and Accountability, courts should clarify the “responsibility, culpability, liability, and accountability” for A/IS prior to the development and deployment. It also states that designers and developers of A/IS should be made aware of the diversity in existing cultural norms among these A/IS.
Awareness of Misuse
EAD states that creators should offer protection against all potential misuses and risks of A/IS in operation. EAD recommends that creators should be made aware of methods of misuse.
It also states that A/IS should be designed in ways that can minimize the opportunity for these systems. Public awareness should be improved surrounding the issues of potential A/IS technology misuse.
EAD states that the creators should specify and operators should adhere to the knowledge and skill required for safe operation. It also mentions that the creators of A/IS should clearly specify the types and levels of knowledge required to understand and operate any given application of A/IS.
Also, creators of A/IS should provide the affected parties with information on the role of the operator and the implications of operator error. Rich and detailed documentation should be made accessible to the experts and the general public.
For more information, check out the official Ethically Aligned Design guidelines