Artificial Intelligence: Synthetic Evolution

As modern society evolves, so does our understanding of our humanity. With the advancement of our abilities forever fuelled by the inherent motivation to expand our potential, we are now questioning the ethics of creating the first artificial intelligence. In that discussion, universal rights will undoubtedly have a place.

Further to the question of humans having the right to create something from nothing, the concern has been noted as to the dangers of creating a synthetic sentience. Rest assured, AI will not happen by accident or by mixing just the right set of chemicals or connecting the right configuration of wires. To understand how fundamentally misguided it would be to fear the sudden discovery of AI, we need to first understand how intelligence occurs naturally.

Intelligence comes from an inherent trait in self reasoning. As any intelligence gathers information, the more it will grow in complexity. As humans, our baseline intelligence does not start from zero. We come with genetic intelligence that provides the right environment for sentience to exist. A child is not born with the abilities of speech, or has complex skills in dexterity. These skills are developed after birth, as we’ve all experienced. What we are given at birth are the tools to develop these skills and capabilities.

So where do these genetic tools come from? Our evolution is defined by the chaining environment our species experience. With each challenge put upon our species comes adaptation equal in response to these challenges in the history of our environment. Understandably, a static environment = static gains in intelligence. Solving additional problems gains additional experience collected in our genetic intelligence.

Learning to walk on two feet — then to run — gained us clear advancements over species who remained in the trees. This genetic intelligence remains with us, and each new born child comes to life with these assets. The more complex we became, the more capable we became. Perhaps far back in our genetic history we straddled the line of sentience — becoming self aware, able to predict potential outcomes from our skills in pattern recognition. The more patterns we recognized, the more our genetics recorded the tools used in recognizing even more complex patterns.

Our genetic intelligence does not include a disposition to speak a particular language, eat only a particular food, or naturally understand how to tie shoes. These are open — allowing us to adapt to unknown variables of an environment specific to each individual. Such specifics are far too varied in the natural timescale of our species; therefore any disposition to perform a particular task must not threaten the survival of a species if the task is dismissed after a few generations. For if it did, humans might have developed a protective fur coat during a 500 year cold snap — only to face extinction when the environment suddenly changed direction.

Likewise, if dogs did not develop a disposition to trust humans, dogs would have remained wild animals. Mutual capabilities in sociality evolved between humans and dogs over a very long period. Trusting humans became part of their genetic makeup — as an instinct, not a hardened rule.

Every instinctive response we have for the most basic situations, were spawned from a repeated task that resulted in a stored adaptation in our genetic makeup. Touch a hot stove, and your instinct is to instantly pull your hand away, no problem solving was needed before acting. The genetic intelligence you were born with prevented you from inaction. The intelligence gained from your experience reenforces the need to maintain the genetically recorded response. If the design of the stove and it’s use remained unchanged for thousands of years, perhaps we would capture the operation of the stove in our genetics — not unlike how pigeons flee from the silhouette of an owl.

Genetic adaptation is also found in e-coli bacteria, where experiments revealed the bacteria’s genes record tools in solving the same problems the same way. Further to the study, it was found that subsequent challenges of similar (but unknown) task were given, resulting in reduced adaptation period — the genetic tools were put to use, making the task easier.

We all start as infants barely able to function above the most basic organisms found in nature. Unlike other species, we cannot stand within hours of birth, nor can we unbury ourselves from a beach, crawl to the ocean and swim away. Given time, complex neural pathways develop to build mental tools in gathering information about our environment and how to effectively interact and adapt to it — all with the help of our genetic intelligence. The complexity of these tools are vast and are only partially charted areas — these areas also include emotions, a very complex toolset.

To suggest AI will spring out of a synthetic soup of chemicals and wires is oblivious to the entirety of how sentient beings gained intelligence. While it’s technically possible to create an AI, it should be understood that a synthetic genetic (preprogrammed) intelligence will be needed to guide the artificial child towards basic learning. Unfortunately, the trial and error we humans endured over thousands if not millions of years is not easily duplicated in a lab or with software — this type of intelligence isn’t free, it must be developed first and has to be just right.

The complexity and effectiveness of any intelligence is proportionate to the genetic intelligence that spawned it. A synthetic intelligence will have to experience the infinite permutations of all possible situations in a simulated environment equal to the entire evolution of the human species —  a very long time indeed. The result cannot come as genetic memories, but as tools — allowing the birth of a living synthetic intelligence, as guided by a synthetic instinct.

If human sentience could not draw from genetic intelligence, our children would likely have been brain-dead at birth, having no ability to breathe much less react to a hot stove. Those who would read that statement and respond “But we would program that in, no problem.” My response would be: While your at it, program a response for every nerve on every millimetre of the synthetic being — as sensory input is a fundamental learning tool — compulsory in gaining self awareness in a spacial environment, and let’s not forget emotions & desires while your at it. Every possible instinct we humans are aware of (and many we’re not) will need to be addressed with the goal of absolute harmony.

No, the world will not be taken over by the discovery of an AI. It will require evolution on a timescale as great as our own. For even as humans, we fall far short of knowing how to deal with our own mental complexities given our clear advancement over a software package. Likewise, to make an airplane, you must first understand how birds fly, and yet, birds are still masters of the sky — taunting us in knowing how to walk and swim all the while.

If AI becomes a reality in our future, be prepared to witness the birth of “dumb”, starting at the bottom of a ladder millions of miles tall.