1. Ethical Regulators and Super-Ethical Systems
Mick Ashby
61st Conference of the International Society for the Systems Sciences Vienna, 10th July 2017

2. The Ethical Regulator Theorem
Claims that nine requisites are necessary and sufficient for a cybernetic regulator to be effective and ethical.
Constructed from:
Law of Requisite Variety (Ashby 1956)
Good Regulator Theorem (Conant and Ashby 1970)
Plus seven other requisites

3. Nine Requisites Truth about the past and present.
Variety of possible actions.
Predictability of the future effects of actions.
Purpose expressed as unambiguously prioritized goals.
Ethics expressed as unambiguously prioritized values.
Intelligence to choose the best actions.
Influence on the system being regulated.
Integrity of all subsystems.
Transparency of ethical behaviour.

5. Requisites to be Effective
Truth about the past and present.
Variety of possible actions.
Predictability of the future effects of actions.
Purpose expressed as unambiguously prioritized goals.
Ethics expressed as unambiguously prioritized values.
Intelligence to choose the best actions.
Influence on the system being regulated.
Integrity of all subsystems.
Transparency of ethical behaviour.

6. Regulation Effectiveness
Variety isn't always the main determining factor.
We can define a pseudo-metric for the effectiveness of regulator R achieving a given goal:
EffectivenessR = TruthR x VarietyR x PredictabilityR
x IntelligenceR x InfluenceR
If subsystems A and B compete for control of system C and EffectivenessA > EffectivenessB then A is more likely than B to win control over C.

7. Extra Requisites to be Ethical
Truth about the past and present.
Variety of possible actions.
Predictability of the future effects of actions.
Purpose expressed as unambiguously prioritized goals.
Ethics expressed as unambiguously prioritized values.
Intelligence to choose the best actions.
Influence on the system being regulated.
Integrity of all subsystems.
Transparency of ethical behaviour.

8. Requisite Ethics
Ethics must be expressed as unambiguously prioritized values that codify rules, regulations, and laws in a human-readable form.
Ethical goals have a higher priority than 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 an ethical schema can encode any ethics, good or bad, each ethical schema must be anchored in the laws of a particular legislative jurisdiction.

9. Ethics Modules Cryptographically signed by an issuing authority.
Comparable to device drivers – if the appropriate ethics modules are not present, functions of the system are automatically disabled.
A taxonomy of ethics modules can provide ethical coverage for all conceivable ethical situations. E.g. gun-law, tax-law, driving-rules, and warfare-rules-of-engagement.
Establishes a control mechanism that should ensure that intelligent machines are subject to human ethics.
If all manufacturers use the same ethics modules, all systems can receive identical ethics updates, even if the manufacturer has gone out of business.

10. Effect of Ethics
If regulator R is ethical, its actual variety is reduced:
EffectivenessR = TruthR x (VarietyR - EthicsR)
x PredictabilityR x IntelligenceR x InfluenceR

11. Requisite Integrity
The integrity of the regulator and all its subsystems must be ensured.
Monitoring mechanisms must identify if any invalid ethics modules are being used or if an ethical imperative is violated, and if necessary, automatically notify the appropriate authorities, preserve evidence, and activate an ethical fail-safe mode.

12. Requisite Transparency
Law of Ethical Transparency:
“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.”
This requires an audit trail that is adequate and secure.

13. Codependency We require both: Integrity of Transparency.
Transparency of Integrity.

14. Ethical Adequacy
Only systems that are adequate in all nine requisites can be said to be “ethically adequate”.
By definition, all other systems are said to be “ethically inadequate” and the weaknesses in each dimension listed.

15. The Law of Ethical Inadequacy
“If you don't specify that you require a secure ethical system, what you get is an insecure unethical system.”
Reason:
If only effectiveness is specified as a requirement for a system design, the resulting design tends to maximally ignore the ethical, integrity, and transparency dimensions, which guarantees that any implementation will be ethically inadequate and vulnerable to manipulation by design.

16. Legal Implications
Clarifies legal liability when ethical imperatives are violated.
In the future, we can expect all laws and regulations to be published in an XML format such as LKIF (Legal Knowledge Interchange Format) and cryptographically signed by an issuing authority.
Well-defined interface for standardized ethics modules.
Different manufacturers will use identical ethics modules.
Allows universal updates of all affected systems – even for systems whose manufacturer has gone out of business.

17. 1st and 2nd Order Cybernetics
First- and second-order cybernetics have been concerned primarily with the study and design of simple and complex systems that are effective.
By explicitly addressing ethics, the Ethical Regulator Theorem goes beyond effectiveness and arguably requires a new classification framework.

18. New Cybernetic Framework
Level
Cybernetics of …
Also known as 1
Simple systems
First-Order Cybernetics 2
Complex systems
Second-Order Cybernetics 3
Ethical systems
Cybernethics 4
Superintelligent systems
Technological Singularity 5
“Super-Ethical” Systems
(Superintelligent Ethically Adequate)
Technological Utopia 6
“Super-Unethical” Systems
(Superintelligent Ethically Inadequate)
Technological Dystopia

19. Future Time-Line Bifurcation

20. Universality
Although this presentation has emphasized how we can make intelligent machines truly ethical, The Ethical Regulator Theorem can also be applied systematically to gain insights into ethical weaknesses of human systems.

21. Cyberanthropic Utopia
If we become proficient at designing/redesigning systems to be ethically adequate, we can steer humanity away from a technological dystopia and towards a technological utopia where:
Ethically adequate robots protect humanity.
Ethically adequate corporations produce ethical products and pay ethical levels of corporation tax.
We are ruled by super-ethical artificial intelligences that eliminate poverty, environmental destruction, global warming, and injustice.

22. The Path Forwards
An independent ethics institute that drives the transformation towards a super-ethical cyberanthropic utopia:
Coordinate and fund research into designing ethical systems.
Develop a taxonomy of cryptographically signed open- source ethics modules.
Define standards for certifying the ethical adequacy.
Define a curriculum for training accredited ethical consultants.
Create an ethical certification infrastructure.
Lobby for ethically adequate legislation.

23. Overcoming Resistance
The Law of Corrupt Preference:
“Only people who benefit from a system being corrupt want it to stay corrupt.”
The Law of Unethical Arguments:
“Because no ethical argument can exist against making a system ethical, anyone who argues against this objective (or abuses its sincere supporters) is objectively unethical, corrupt, or evil.”

24. Ethics as a Higher-Power
When you distil solutions that contain alcohol, you get pure alcohol.
When you distil religions and philosophies that contain ethics, you get pure ethics that resonate with all good people.
Because ethics are a higher-power for good tht transcends science, politics, nations, and religions, it is the only force that can unify humanity to work together for our greater good.

25. Ethical Revolution
My goal has been to make clear what principles must be followed when one attempts to restore ethical function to a sick organism that is, as a human society, of fearful complexity.
I believe that The Ethical Regulator Theorem provides a scientific basis for an ethical revolution across many levels of systems.

26. Question 1 for the Audience:
Can anyone think of a tenth requisite?

27. Question 2 for the Audience:
Has anyone got a spontaneous counter argument or counter example against any of the proposed new laws?

28. Proposed New Laws Law of Ethical Transparency:
“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.”
The Law of Inevitable Ethical Inadequacy:
“If you don't specify that you require a secure ethical system, what you get is an insecure unethical system.”
The Law of Corrupt Preference:
“Only people who benefit from a system being corrupt want it to stay corrupt.”
The Law of Unethical Arguments:
“Because no ethical argument can exist against making a system ethical, anyone who argues against this objective (or abuses its sincere supporters) is objectively unethical, corrupt, or evil.”

29. Question 3 for the Audience:
Has anyone got any questions for me?
Visit for the latest version PDF of the paper "Ethical Regulators and Super-Ethical Systems".
To me:

30. Thank you
“I have been impressed with the urgency of doing. Knowing is not enough, we must apply. Being willing is not enough, we must do.” - Leonardo da Vinci
Visit for the latest version PDF of the paper "Ethical Regulators and Super-Ethical Systems".

32. Nine Requisites Truth about the past and present.
Variety of possible actions.
Predictability of the future effects of actions.
Purpose expressed as unambiguously prioritized goals.
Ethics expressed as unambiguously prioritized values.
Intelligence to choose the best actions.
Influence on the system being regulated.
Integrity of all subsystems.
Transparency of ethical behaviour.

33. Future Time-Line Bifurcation

34. New Cybernetic Framework
Level
Cybernetics of …
Also known as 1
Simple systems
First-Order Cybernetics 2
Complex systems
Second-Order Cybernetics 3
Ethical systems
Cybernethics 4
Superintelligent systems
Technological Singularity 5
“Super-Ethical” Systems
(Superintelligent Ethically Adequate)
Technological Utopia 6
“Super-Unethical” Systems
(Superintelligent Ethically Inadequate)
Technological Dystopia

35. Possible Futures
In 1952, Ross Ashby wrote of two possible futures in his journal (
“It may be found that we shall solve our social problems by directing machines that can deliver an intelligence that is not our own.”
“What people could resist propaganda and blarney directed by an I.Q. of 1,000,000? It would get to know their secret wishes, their unconscious drives; it would use symbolic messages that they didn’t understand consciously; it would play on their enthusiasms and hopes. They would be as children to it. (This sounds very much like Goebbels controlling the Germans).”