Artificial Intelligence and Consciousness

Welcome to my new blog.  About once a month, I'll be posting commentary, speculations, criticisms, remarks, and brief book reviews about science fiction and related things.  I'll be posting my own science fiction stories, less often, under the "Stories" heading on the home page.  Please comment!

In my science fiction story, In Defense of Biohumans, posted today, I speculate about a powerful, fully conscious artificial intelligence (AI) and how it might be absorbed into society, or vice versa.  I don’t want to tell you too much.  I hope you will read and enjoy the story.  It’s on its own page under Stories on my home page.  Just click. 
 

There’s been a lot of scare language lately about the dangers of AI.  Some of the worst is about AI getting out of control and wreaking terror and devastation either through a colossal mistake or through conscious intention to advance its own program.  Can an AI ever become conscious?  Philosophers western and eastern have clenched their brains for centuries over the question of whether our consciousness (or mind) arises from matter or is a special something else.¹ They’ve yet to resolve the question. This question could, of course, be of critical importance in deciding whether a machine can become conscious.
 

One approach to this question that could be aided by scientific tools is to trace how the functions of our own conscious minds may have developed naturally from physical and chemical processes that produced life on earth, and then compare the features produced by such a development to the peculiarities of our own consciousness as a test of plausibility.  This is the approach taken in a recent book by anthropologist and biologist Terrence Deacon, Incomplete Nature – How Mind Emerged from Matter (W.W. Norton 2012) which provides a remarkable discussion of how consciousness may have emerged step-by-step from known physical processes.  Deacon shows how biochemical processes in ancient earth conditions could have driven early life forms to evolve consciousness.  He doesn’t conclude that he’s proven the case, but rather that the mechanisms he unfolds could form the basis of a program to investigate and prove or disprove hypotheses for how our consciousness emerged from matter.  This program, if carried out, could go beyond the basic question of “Could it happen?” and help in developing AI, since the more we understand about human or animal consciousness, the features that make it peculiarly human among other things, the more we can understand about how a machine consciousness might occur.  And how it might be different from the human variety.
   

Although Deacon doesn’t discuss AI in his book, his arguments lend credence to the idea that consciousness can emerge from known physics and chemistry. He sets forth a very orderly exposition of ideas and facts that lead to plausible hypotheses regarding how consciousness emerges from the physical and chemical processes of biology.  It’s a big book, 545 pages plus a glossary, notes and references. It’s enjoyable reading though, because he takes care to provide the necessary information in clear prose so that an attentive reader not familiar with the underlying science can follow it. 

Taking a completely fresh approach to the worn discussion about the nature of consciousness, Deacon invents several important new terms to explain his ideas, beginning with “absential,” which he describes as the “elusive character” of many things associated with life that depend on what is not there, such as appetite, aspiration, desire, and more general functions such as references, purposes and values.  He says this absential quality is “a defining property of life and mind.”  Later he introduces “ententional” phenomena, which “have an orientation to a specific constitutive absence, a particular and precise something that is their critical defining attribute.”  This is distinguished as a more general form of and precedent to “intentional” which is a property of a conscious mind.  Each entention has a corresponding telos, or end. 

After an entertaining side discussion on the historical fallacies of various forms of homunculi as models of consciousness, there is a discussion of teleonomy (end-seeking as law of nature) and various writings about it in natural systems.  This leads to discussion of general tendencies in nature especially entropy and homeostatic feedback systems, natural and artificial.  Things that just go along, ruled by these universal tendencies are “orthograde” and those that don’t are “retrograde.”  Life is retrograde.  Then another side discussion on writings about “emergentism” which posit life and other processes are higher order and emergent from lower order physical phenomena.  Deacon introduces the idea that constraints form in causal relationships with processes that are “supervenient” on lower order processes, and these constraints channel the lower order processes in ways that can be teleonomic and ententional, and then intentional.  So far, it’s an abstract, philosophical-sounding discussion, the main point being that you can get from physical and chemical drivers (which may set up a pattern that improves a system's chances of persisting) to intentions, the first hint of consciousness.  My summary above is highly condensed. Deacon’s of course is much, much longer but more convincing as well as entertaining. 

Deacon describes in chemical detail hypothetical pre-life configurations of organic chemicals that can form reciprocally autocatalytic, i.e. self-sustaining chemical systems that also, and critically, form containing capsules.  With increasing complexity, these systems develop informational codes (e.g. DNA) that allow passing developmental and structural information to further reproductions of themselves.  Of importance here is that the DNA is secondary.  The system must first be self-reproducing, before a genetic code is added to improve the reproductive process.  The original self-reproducing system is necessarily simpler than DNA and relies on autocatalytic and thermodynamic phenomena to begin the self-reproducing process.  There is a beautiful side discussion of “morphogenesis” as an intermediate step in this process, where certain naturally occurring physical processes (described by Deacon in detail) produce unanticipated physical shapes and forms (illustrated graphically), which could be similar to the primitive cellular forms that contained the original reciprocally autocatalytic systems.   Basically, he explains how the persistent systems and structures of life could emerge from the chemistry and thermodynamics available on the early earth.
   

Deacon introduces the concept of “teleodynamics,” where the evolution of self-contained self-reproducing systems produces an “end-directed” system that continually enhances the viability of the system through feedback processes.  He describes teleodynamic systems as supervenient on morphodynamic systems which are in turn supervenient on homeodynamic systems.  Whew. This hierarchy might seem like so much philosophy, were it not for Deacon’s wonderful descriptions and graphics of real physical and biochemical systems that demonstrate the principles of each stage. The hierarchy results from those physical and chemical processes operating in accord with thermodynamic principles, rather than some magically “emergent” process that forms and then organizes the lower order processes.  What we have today as living cells are thus produced from a “bottom-up” process rather than top-down from an original replicating DNA or other master molecule. 
 

Deacon then proceeds to the importance of information to the viability of the autogenic system: “Evolution generates the capacity to interpret something as information.”  By developing persistent (inheritable) constraints in the system, information is increasingly used to govern the behavior of the system.  This ongoing process of information-seeking and system self-modification is the basis of consciousness.  Consciousness is information management driven by the need to thrive.  “…the experience of being sentient is what it feels like to be evolution.” (p. 502).  Neurons operate in a statistical fashion to provide the function. 

The implication of Deacon's book is that the self-preservation drive of evolution is what brings consciousness into being in living things.  Consciousness enhances survival.  Part of its self-preservation effect is that it creates in the organism a clearly defined sense of self vs. outside self, which is supported by the specific physical and chemical configurations of the earliest self-reproducing systems that developed on earth.  I wonder whether machine consciousness might be a bit different in this regard, leaning more toward absorption of as much non-self as possible rather than creating effective barriers. 
 

Deacon’s step-by-step exposition of a plausible explanation of the origins of both life as we know it and consciousness is amply referenced to both physical and biological facts and philosophical thinking of others.  Nevertheless, he presents it not as a conclusion but as a hypothetical framework to guide investigations of the origins of life and consciousness.  And maybe also the potential for machine consciousness. 

---

¹ See Lucretius, De Rerum Natura Book 2 line 886 and ff (“from insensate germs, the sensible is gendered…”), and Chinese Philosopher Zhu Xi, 12^th century AD as quoted in Joseph Needham, The Grand Titration - Science and Society in East and West 1969): “The mind’s function is perfectly natural, something which matter has the potentiality of producing when it has formed itself into collocations with a sufficiently high degree of pattern and organization.”