Artificial Intelligence (AI) has exploded into the headlines worldwide.
Thrilling promise…chilling threat. This development is going to be epic — and we need to get it right, or else.
In the same way that I believe humanity has been dangerously missing a Universal Morality to guide our self-organizing, so is AI missing this critical “true North.” The UN, the WEF, the White House, multiple governments and others have issued their versions of an “AI Bill of Rights,” but they are inadequate, self-serving covers for more centralized power and control.
I was invited by Matthew James Bailey and the folks at AIEthics.World to offer guidance in creating a Declaration of Ethics for Artificial Intelligence. It is intended to serve like the Declaration of Independence or the Ten Commandments. I shaped the original and incorporated excellent suggestions by dozens of the people I most respect in the fields of freedom, privacy and ethics.
This offering is a work-in-progress that hopefully will continue to improve and help guide us into a thriving future.
The computer era has made much of life more efficient and more organized for humankind.
The next step in this realm has been the introduction of Artificial Intelligence, or AI — computers and robotic entities that can perform many human tasks, including learning and decision making. To date, we have seen the benefits of this in situations such as assembly lines and medical procedures, and we are now on the cusp of that same technology making possible the production of robots for the purpose of controlling society.
We are at risk of seeing a loss of human freedom and even a loss of what makes us human, like genuine compassion, wonder, vision, intuition and love.
While AI can be helpful, it is, by definition and by the intent of its creators, capable of developing beyond the control of its original programmers. This capacity to gain skills outside of those originally programmed and beyond those of humans is in fact the goal of many of its primary funders and incubators. This is called “The Singularity.” Along with the limitations of human consciousness and biased values that program the AI, this leaves open the question as to whether a code of ethics, a “moral compass,” will be instituted that will control the algorithms and robots. If we are to enjoy the benefits of AI while eliminating or minimizing the risks, we need clarity on and a means to enforce such a code.
This Declaration of Ethical AI is intended to establish that code.
Ethical behavior by humankind is based upon the non-initiation of force or fraud toward others and their property. This is known as the Non-Aggression Principle (NAP). We assert that it is essential that no robot be created without this principle being programmed as a central guide to its function — regardless of place, time, culture, community, gender, race, age or condition.
Vital Distinctions Underlying This Declaration
- Morality is the practical day-to-day application of universal and rational ethical theories. It is analogous to the relationship between engineering and physics.
- Coherent ethical guidelines cannot impose positive (pro-active) moral obligations upon people (i.e. THOU SHALT, as opposed to THOU SHALT NOT), because that requires coercion, which is a violation of the NAP. Thus, rational ethics can only be protective, i.e. outlining boundaries that can be justly enforced, rights that can be morally defended. This starts with the recognition that we each own our own bodies, the fruits of our labor, and our rightfully gained property.
- Rightfully gained property is created or acquired through original development (like homesteading unowned property or original intellectual property), purchase, trade, gift or inheritance.
- Positive (proactive) moral virtues, such as courage, generosity, friendliness, compassion and humor can be promoted, encouraged and modeled, but they cannot be imposed or enforced. Positive moral traits and behavior are only a value when they are freely chosen.
- Honoring differences in culture, religious beliefs, sexual preferences etc. is ethical only as long as the NAP is upheld and individuals are not being violated.
- Many so-called “Transhumanists” dream of turning over planet Earth to robots and computers. They are beginning by merging digital devices with our organic biology. They have struggled unsuccessfully to get consciousness and life from digital calculation. Ethical AI can help lift our quality of life but it can never and should never replace our sacred purpose to evolve our living consciousness, learn how to love one another, steward our natural environment and live in harmony with the cosmos that is our home.
Examples of What We Cannot Ethically Have in Artificial Intelligence
- Covert or deceptive terms of service. (Fraud)
- Theft of personal data. (Stealing)
- Withholding of information that is critical to informed consent. (Fraud)
- Forced, mandated or covert implants, injections or aerosolization of nanobots. (Invasion of body)
- Programming AI to violate the Non-Aggression Principle in any way. (Automated immorality has the capacity to affect many billions of people without recourse or consequence and has no consciousness to motivate its reflection or change.)
- Anything that imposes an unknown or unacceptable risk for those potentially impacted; for example (but not necessarily limited to): contamination or destruction by nuclear radiation, gain of function bioweapons etc. (Assault / murder)
- Censorship of free speech (Theft of bodily autonomy, since the products of one’s labor — words — are eliminated without permission. (The courts must determine the limits of speech, such as incitement to violence, slander, libel etc.)
- Poisoning of air, water, food, people. (Assault / murder)
Declaration of Ethics for Artificial Intelligence
The Non-Aggression Principle (NAP), the core of all human ethics, recognizes that no individual human has the right to initiate force, coercion, or fraud against another — except in true self-defense. Artificial Intelligence (AI) must be subject to the same code of ethics as humankind itself.
Such restrictions on Artificial Intelligence shall include, but not necessarily be limited to:
- Destruction of property (including data and intellectual property)
- Violation of Informed Consent
All the above protections are understood under this Declaration to be the natural and equal rights of all individual humans, regardless of race, gender, religion, nationality or ethnicity. All AI shall notify every individual if anything is being done to them contrary to their stated will. All AI programs and robots are to be created and applied in accordance with this Declaration.