Toward a Precise Definition of a Universal Ethical Principle
Ben Goertzel
February 20, 2004
In my recent essay Encouraging a Positive Transcension (Goertzel, 2004a), I proposed an ethical principle of “Voluntary Joyous Growth” – which holds basically that a good ethical system is one that tries to balance the three factors of choice, joy and growth. Defining these three concepts is not easy – they’re all a bit slippery, and each one conceals a vast mass of ambiguity, subtlety and human history. This brief note indicates my line of thinking regarding the definition of these concepts in terms of (human and AI) psychology.
I note that, in order to impart an ethic to another mind (AI, human, whatever), explicit definitions are of limited use. Examples are more important, and most important of all are the collective exploration of scenarios – living and working through the ethic in “real life.” But nevertheless, explicit definitions can be of some value in guiding these more practical and essential aspects of ethics.
Growth is the simplest of the three basic principles. I conceive growth as the creation of more and more pattern. If one defines a simplicity measure, one can mathematically define the “intensity” of patterns (Goertzel, 1997), thus quantifying what it means to have more and more pattern. Furthermore, if one accepts the theory of consciousness I’ve presented in (Goertzel, 2004b), then more intense patterns correspond to more intense qualia, so that more patterns means more experience. So growth means that the universe becomes more aware.
Choice is a little more complicated: choice, as I analyze it in (Goertzel, 2004c), has to do with the maintenance within intelligent systems of a “virtual multiverse model” of the universe, in which multiple potential future universes are studied, and this study is used to guide system dynamics. Valuing choice means valuing systems that guide their actions using internal virtual multiverse models. In essence, then, choice means valuing the explicit embodiment of the multiverse within the universe. Valuing choice means valuing virtual multiverse creation.
In brief: More pattern is good, more experience is good, and multiplicity of possibility is good.
Choice helps with growth, because systems that embody virtual multiverses tend to be good at generating new patterns and experiences. Growth helps with choice, because it takes fairly complex patterned systems to spawn internal virtual multiverses.
Joy – as I analyze it in my essay on emotions (Goertzel, 2004d) – is what happens when a system finds itself overwhelmed with system-wide response patterns that occur in reaction to the successful achievement of its goals. More specifically, what I refer to there as “spiritual joy” is what happens when a system is overwhelmed with a complex dynamical pattern that embodies harmony between the inside of the system and the world in which the system is embedded.
Valuing joy means valuing systems setting goals and achieving them. Valuing spiritual joy means valuing the quest for harmony between self and universe.
Achieving goals is good; harmony with the universe is good.
Successful goal-achievement, in combination with the values of choice and growth, means that the values of choice and growth are propagated through the various systems of the universe via their goal-systems.
Spiritual joy is a reflection of the basic dynamic by which the universe seeks to overcome the paradox of one-versus-many (for a poetic exploration of this theme, see Goertzel, 1998). All is one, yet each thing is separate and distinct – this paradox lies at the heart of being, and one view of our ultimate purpose is to overcome this distinction and completely manifest both our unique separateness and our oneness at all times. Seeking harmony between the interior and exterior worlds is the way to fufill this purpose.
One point to note is that, in my perspective, all of these three values are defined in terms of the concept of “pattern” – but “pattern” is not an objective concept. The mathematical theory of pattern defines pattern in terms of a more elemental concept of “simplicity.” Thus, judgments of the amount of choice, growth or joy in the world are ultimately dependent on the simplicity measure implicit in the judging mind. This leads to an important point regarding the creation of nonhuman AI’s valuing growth, choice and joy. If these AI’s measure simplicity very differently from humans, they will perceive different things as growing, joyful and free. Of course, it’s not viable to restrict a superhuman AI to have the same simplicity measure as humans – but at very least, one can ensure that one’s AI has a simplicity measure that’s inclusive of human assessments of simplicity, in the sense that whatever humans perceive as simple, so does it. This means that the AI will see choice, joy and growth where humans do – but may also see it in other places.
Finally, a cautionary note is in order: Of course, I don’t claim to have fully explicated any of my three basic values – joy, growth or choice – but I hope to have elucidated a little bit of what lies inside them. There is much more to discover.
The following references are to essays posted on the Dynamical Psychology e-journal, at http://www.goertzel.org/dynapsyc/