Ethical Regulators and Super-Ethical Systems
Keywords:ethics, transparency, robotics, singularity, cyberanthropic utopia
The Good Regulator Theorem proved that every effective regulator of a system must be a model of that system, and the Law of Requisite Variety dictates the range of responses that an effective regulator must be capable of. However, having an internal model and a sufficient range of responses is insufficient to ensure effective regulation, let alone ethical regulation. And whereas being effective does not require being optimal, being ethical is absolute with respect to a particular ethical schema.
This paper takes the Good Regulator Theorem, and unifies it with the Law of Requisite Variety and seven other requisites. The resulting Ethical Regulator Theorem has implications for designing and certifying explicitly ethical systems. It claims that the following nine requisites are necessary and sufficient for a cybernetic regulator to be effective and ethical:
Truth is not just about information that the regulator receives as inputs or treats as facts, but also the reliability of any interpretations of such information. If the regulator’s information sources or interpretations are unreliable, and cannot be error-corrected, then the integrity of the system is in danger. And if the perceptions of the regulator can be manipulated, it can be tricked into making decisions that are ineffective or unethical.
Variety in the range of possible actions must be as rich as the range of potential disturbances or situations. This is The Law of Requisite Variety.
Predictability requires a model that can be used to select the actions that will give the best outcome. This is the Good Regulator Theorem.
Purpose is expressed as unambiguously prioritized goals.
Ethics are expressed as unambiguously prioritized values that have a higher priority than the goals for purpose. By always obeying the relevant highest priority ethical imperatives, the regulator is guaranteed to act ethically within the scope of the ethical schema. Because ethical schemas vary between legislative jurisdictions, they are handled as plug-ins.
Intelligence must be applied to the previous five requisite types of information to select the most rational and effective ethical action from the set of possible actions.
Influence is the existence of pathways to transmit the effects of the selected actions to the regulated system. This is not a property of the regulator itself, but a function of the connectivity relationships that span from the regulator’s outputs to elements of the regulated system and its environment.
Integrity of the regulator and all its subsystems must be assured. Monitoring mechanisms must identify if an ethical imperative is violated and, if necessary, automatically notify the appropriate authorities, preserve evidence, and activate an ethical fail-safe mode.
Transparency is defined by the Law of Ethical Transparency, which states “For a system to be truly ethical, it must be possible to prove retrospectively that it acted ethically with respect to the appropriate ethical schema.”
Integrity and Transparency are codependent because we require integrity of transparency, and transparency of integrity.
Because this theorem is independent of the ethics schema that is used, it provides a basis for systematically evaluating the adequacy of existing or proposed designs for systems that make decisions that can have ethical consequences; regardless of whether the systems are human, machines, or cyberanthropic hybrids.
In addition, a new framework is proposed for classifying cybernetic systems, which highlights the existence of a possibility-space bifurcation in our future time-line, and the implementation of “super-ethical” systems is identified as an urgent moral imperative for the human race to avoid a technological dystopia. Concrete actions are proposed to steer our future towards a cyberanthropic utopia.