THE COMING AGE OF A.I.: UNDERSTANDING COGNITION
© Copyright 2003 Video Gamer X Publications

In my previous article, I introduced A.I., nanotechnology, and cybernetics, all technologies that are currently being developed and experimented with around the world. One thing that must be recognized is how these technologies will be implemented and when. It is inevitable that computer intelligence and human intelligence will intersect. It's not a matter of if, but when this will happen. Another branch of evolution is taking place right now, and a second intelligent species is about to be created here on planet Earth. But in order to understand the mind of this new life form an examination of the workings and processes of our own computer, the human brain, must be undertaken.

The human brain is capable of what has been approximated theoretically at between 100 Trillion and 100 Quadrillion operations per second. It contains about 100 billion neurons, with each synapse in the neighborhood of 1000 interconnections. Mind you I doubt every single synapse in the human brain is being using simultaneously but no doubt that a lot of raw data processing power is required to keep you alive, conscious of the world around you, and capable of learning and reasoning. Neurons are the biological building blocks of data storage and transfer, in essence a combination of RAM, ROM, and microprocessor logic gates all in one.

Neurons work by sending an electrochemical pulse along axons to neighboring neurons. Pathways are generated between clusters of neurons, and it would seem that the shape and location of those neurons in the human brain determine things like memory.

The fastest computer in the world, NEC Earth Simulator, is currently located in Japan; it is capable of performing around 36 Trillion Operations per second at peak capacity. That is still only a fraction of the capability of the human brain, however time and the advance of technology does not remain stagnant. If you know anything about computers or read articles on the internet in technology news sections and message forums, you may have heard of a frequently referred to term called, "Moore's Law." Moore's law states that for every 18 months of time, the power of microprocessors doubles. This is most easily observed if you just reflect on the speed of your first computer compared to the one you currently own. I have a personal example. In late 1997 I owned a PC with a 233 MHz processor, 6 GB Hard Drive, and 64 MB of RAM. I spent about $1,500 on this computer. In late 2002 I bought the parts and assembled a computer myself, the one I am using currently. This computer has a 2,530 MHz Processor, 200 GB Hard Drive capacity, and 1 GB of RAM. I obtained the parts for around $1,300, taking into consideration that I added unnecessary cosmetic upgrades like colored lights and cooling fans and a svelte black and gold Plexiglas case. During those 5 Years my old computer's total capacity and level of efficiency was dwarfed by a scale factor of around 10 times. Currently, as I am writing this article in mid 2003, personal computer microprocessors from Intel clock in at about 3,066 MHz. See where this is going?

Let us assume that we use the timescale of today and say that NEC's Earth Simulator is the fastest computer on Earth in 2003. Using Moore's law, which is has proven entirely true since he first mentioned over two decades ago it is possible to give a good guess at where computational capacity be in the future. For example in 2005 there should be a supercomputer capable of around 70 TeraFLOPS. (1 Teraflop is saying 1,000,000,000,000 floating point operations per second) Mid to late 2006 a computer may appear with 140 TeraFLOPS, 2008 - 300 TFLOPS, 2010 - 550 TFLOPS, 2012 - 1200 TFLOPS (1.2 PetaFLOPS), around 2020, there will be a computer that performs 10,000 TeraFLOPS, or somewhere in the neighborhood of the human brain. If you project the development of computers over time you should arrive at the intersection point where computers and the human brain have equal efficiency and capacity and at this intersection point it's a safe determination that self aware artificial life will be created. This of course is just talking about one specific computer arrangement or supercomputer, rather than a concept like "Matrix computing" or "Cell Computing" This would allow the same calculations per second only spread out between millions of smaller computers. Right now, if all the home PC's on the planet could work together sharing processing power, they could probably reach the levels of computation I was talking about in the future, however PC's are used for other tasks but there are pieces of software like Seti@Home that work on this Matrix Computing concept.

If a major breakthrough in technologies like Quantum Computers, Neural Computers, or DNA computers occur then things could happen a lot sooner and the Moore's law timetable could be rescaled. Although unproven and an aspect of metaphysical study, the conclusion of the Mayan calendar occurs on December 21, 2012. Is that the date humans give birth to "A.I. consciousness?" We'll just have to wait this one out and find out for sure, but realistically speaking, such massive computing power of the future can only reasonably be used for one thing, artificial intelligence. It was human brains belonging to visionaries like Albert Einstein, Thomas Edison, and Stephen W. Hawking that made rapid advancements in science possible during the 20th century. Imagine the scenario of such massive computational ability being used to discover more about the universe that we still do not understand. It's entirely probable that shortly after computers surpass human cognition that a new era for humanity will occur where Sci-Fi technologies like antigravity, folding space, time travel, and the perception of higher dimensional existence will occur.

In order to understand what a computer must do in order to duplicate the human brain the distinct aspects of being "alive and aware" must be defined. What makes you "you" is not just a mass of neurons but also the programming that is used by the brain. If A.I. were to take a physical form, it would have to have sensory organs, or input devices. We use our eyes, like a stereo camera to view the world, this is how visual data is acquired and relayed to the brain. We see colors based on the wavelength of light color that is reflected to our retina. There are small features called cones and rods located at the focal point of the cornea on the back of the retina that detect color, and pass this along the optical nerve to the visual cortex. Once there the brain obviously uses this data and processes it for something. For example, your eye sees a room, in the room your shirt is lying on the floor, with that data, you refer to the memory of your mother chastising you for leaving your clothes on the floor, your brain tells your legs to move, you pick up the shirt, fold it, and put it in the dresser drawer. Your brain processes the emotional response of "relief" to the stimulus of putting it away and recognizing that it doesn't have to endure another unpleasant experience of your mother complaining. This chain of events demonstrates how various events and stimuli interact with memory and emotion. Our ability to detect sound augments our vision. Our ears are specially designed input devices shaped to funnel sound to a vibrating surface that is connected to small bones, which are connected sensory nerves that get relayed to brain. If people couldn't hear, they wouldn't know if someone was walking up from behind. In nature this is necessary for survival to hear predators that otherwise might be hidden in tall grass or behind some rocks. Just like data obtained visually, sound is recorded and identified from memory. If you hear a cat meow, but can't see the cat, you will assume it's a cat because this is the proper association of that sound with that object. If you've never heard the sound before and it is similar to something you've heard then your brain will approximate and deduce that it is similar to "such and such" data from memory.

The human ability to smell is not as pronounced as many Animals in nature, but nevertheless provides another tier of perception needed to survive. More than likely the purpose for this sense in evolution is to identify potentially hazardous materials or avoid infection with disease. It also augments the ability to taste by associating an aroma with something consumable. The nasal cavity is shaped to pass air across a sensory zone as it enters the lungs. Odors can trigger memory identification of a visually acquired image just as sound can. If you smell "Pledge" you will think of the can of Pledge, possibly the sound of it spraying out, and the last time you used it. If you smell feces you know that something has defecated somewhere, and the natural instinctive inclination is to either get away from the offensive odor, or to remove the source of the odor. If you smell a "barbecued hamburger on the grill" you will remember the last time that you ate one of these, as well as the texture of the meat and bun and the stimulation of pleasure you had while eating it due to the high fat and salt content. You might even vocalize the words, "Yum" or "I'm hungry."

The sense of smell and taste can work independently but most often go hand and hand. In order for humans to survive it is required that we ingest nourishment. Our biology is set up so that this region of sense is located at the only opening that leads to the stomach and digestive system. Ages before there were grocery stores and Pizza Huts, people gathered food by any means necessary. Instinctively it seems that sugary, salty, and fatty foods are the most desirable to the brain and body. Foods high in carbohydrates and fats provide the most energy for the body and these most often taste the best. In nature these are found in fruits and meat, but in our modern society humanity has enumerated the possible combinations of flavor and nutritional value (or lack thereof) in this regard. When the body is deficient in a nutritional need or pleasure center stimulus it may exert a "craving" that triggers a memory related to flavor and related sensations of something being ingested. Simply put, if you are thirsty, you think, "drink." A physical A.I. body may not require this kind of nourishment, so the necessity of taste as an input device may not merit its inclusion. However, an A.I.'s physical body might have certain needs in order to remain alive. If a requirement dictated that it plug into an electrical source to recharge a battery or power source this might trigger the same feelings and pleasure or reward obtained from eating. This might have to be programmed or may naturally result as an evolutionary trait after A.I. replication.

Another key input device sense that is a component of human perception is the ability to determine the texture of objects in our environment. By nature of just existing requires us to touch everything in our surroundings. From the moment you grow skin and tactile nerve cells you are aware of the world around you. If you were blind, and deaf, you would still know that you occupy some kind of world where there are surfaces, shapes, weights, and textures. Our bodies are covered in tactile neurons that relay information directly to the brain. These are strategically located to optimize their distribution over the entire exterior surface of our body. Just like the other senses the brain relates touch to previously recorded memory. Objects or environments that cause stress or discomfort to the body trigger pain, pressure, or temperature, which are self-preservation responses to stimuli. If you walk outside without sufficient clothes on during a snowstorm your nerves in your skin will process the temperature change, simultaneously relay this data, and the brain will instinctively recognize the threat to and register discomfort. The same holds true to objects that cause physical damage to the body. If you cut yourself you feel pain because it is damaging your body, and the brain preservation response is to get away from the object or environment that is harmful. Touch sensitivity goes beyond just a preservation response. It provides additional data regarding things detected by other senses, most predominantly visual. An association is made between the visual appearance of an object, and the physical weight and texture. It's a very powerful input device capable of determining minute subtleties otherwise unavailable to the other senses. For example one plant's leaf could be placed next to a synthetically created one that has the exact same color, shape, and appearance, might even make the same sound when it falls on the ground or table. You would feel the difference between biological matter and synthetic material even though your visual cortex was processing the two as identical. An A.I.'s "skin" must be permeated with tactile sensors, it would have to register similar response to stimuli that humans do. If a temperature or physical encumbrance was beyond the acceptable tolerance of the being's body it would have to process a string of responses and decisions in order to escape from this situation, a preservation response. It would also need this data in order to determine the amount of applicable pressure required to complete various tasks. Negotiating its surroundings would require the ability to touch. I imagine when A.I. replicates it will determine the most efficient physical shape to survive and interact with the world around it.

Creativity seems to emanate from another region of the brain in which memory is associated. It exists near subconsciousness at the edge of dreams but close to logic. When a person spontaneously generates information, thoughts, or ideas, there is indication that this information is based on what has been previously acquired, experienced, and stored by the brain. A random thought, is not entirely that random, it is based on an abstract extrapolation of all the things you already know or other concepts that you have pondered before. Creativity and innovation can be a byproduct of cognitive reasoning and an internal process of forming a hypothesis based upon previously observed and acquired information. Free will also plays a part by reaching out into unknown territory exploring it and reporting back it's findings. These things may all happen within a fraction of a second when a choice is made. For instance, while an artist paints an image that he has inside his mind's eye or of something observable in reality, he may make thousands of processed decisions on how much of each color pigment to add to other colors, how much pressure to apply on the brush to make a stroke, and what shape and type of stroke to make. A beginning artist will not have as much skill as one who is practiced because the beginner must rely on trial and error until the most accurate or efficient end is obtained to the means he is searching for. Talent resides where the painter or creator can learn and correct error ratios very quickly as well as extrapolate new thoughts efficiently and profusely. Furthermore curiosity and the innate human desire to gain more understanding about our environment and the reality we occupy, spawns a craving for knowledge and drives forward the processes of thought and creativity.

Can an artificial mind be creative? Based upon the set domain surrounding memory and experience, cause-effect, trial and error, and free will this could be called a kind of program or system of processing and creating new data. The process for spontaneous thought generation could result as a natural byproduct of the other capabilities and reasoning skills. An A.I would have to be innately curious like a baby and a child is about the world around it, and have an unquenchable thirst and capacity for new data. The more data it acquired the more apt it would be to be creative and self-inspiring, just like a human being. One important piece of learning that has forwarded human evolution and the development of society and culture is that humans are very social creatures, and often transfer thoughts and ideas from one to another. We use speech to do this. Words, images, symbols, and sounds are used to represent concepts, objects, ideas, thoughts, etc. and through language we can share information readily between compatible languages. Without communication data can not be passed from one entity to another. Communication augments the brain and works on a similar principle to Matrix/Cell Computers, where if a piece of information or thought is generated by one it can be shared with the rest of the individual units, where in turn those units can process it, and generate separate thoughts that can be shared. An artificial mind would have to be either able to communicate with other humans effectively or communicate with other artificial minds. However, the time it takes to share thoughts through spoken language is far less efficient than the speed at which computers can transfer data and information. It's a good bet that if Artificial Intelligence has language it will be entirely digital in nature, and more than likely what would be best associated with a kind of telepathy. Unlike the inefficient ways in which humans have to gather and share thoughts from other humans, AI's communicating in the same language or format would be able to relay their information virtually instantaneously without any deterioration in data. This is essentially the notion behind a hive mind, where each unit or individual is part of a shared pool or collective of information that forms a macromind and is an extension of the Matrix/Cell structure. Humanity is essentially a hive mind but a very inefficient one where trillions of data groups and thoughts can be isolated and fractured due to an inability to instantly communicate the information to the whole species. The diversity of humanity on Earth is a good example of not everyone having the same information at the same time. Imagine if someone else was sitting in class learning Calculus and you could tap into their brain and learn the same thing they learned without having to be there, this is the nature of an efficient hive mind. Technology could augment human communication with "mind modems" where we gain the ability to share thoughts without visual or aural communication.

Another component of human learning that AI must accomplish is the understanding of common sense knowledge, or unofficial rules that govern the parameters of our perception of reality. For instance, we know that "all dropped objects fall down" and that "when you say 'Hi" to someone, the receiver of the greeting is usually the one perpendicular to the plane of the face, directly in front of the speaker. Also, simple concepts like, "three identically sized decks of cards can be stacked on top of each other" is something that an AI would have to either learn on its own through trail and error if it had limbs capable of grasping objects or be programmed with. More abstract concepts would be something that an AI would have to be taught; most often linguistic idioms and clich�'s that if taken literally would make no sense at all mean something entirely different. For example, "Quit, beating around the bush, and get to the point!" In literal terms you would expect to see someone beating something around a bush; however in American culture we commonly know this means someone is avoiding the direct issue of discussion at hand. Now with a system of intercommunicative AI minds, learned data could be shared instantly with the other members of the network and thus all members would then become that much more intuitive. I imagine that if AI becomes self aware, it will discard the inefficiency of human language and choose some kind of mathematical association and classification of information. In essence it will create its own language to program and continually upgrade itself.

COMING SOON:

Part 3: The Ethics of AI

Part 4: Does AI Have Soul?

BACK TO THE CROSSROADS