Tag Archives: founding fathers
Darkly Wise and Rudely Great
Imagine you are a great historical scientist and discover a disturbing truth about nature. Perhaps you discover that the world is not flat like a map, but in fact round like a tomato. Or, maybe you discover that instead of the Earth having a special place at the center of the universe, it is a just an ordinary rock floating in space – distasteful and troubling revelations.
The naturalist, Charles Darwin, was in such a situation in 1859. Although a devout Christian, Darwin’s pursuits led him to discover the mechanism by which life on Earth developed from a common ancestor, leading to the obvious conclusion that humans evolved from their closest relatives, the great apes. It was the most disgusting idea of the century, and he was personally troubled by it for the rest of his life.
The path Darwin’s life took made him a likely candidate to discover the mechanism of evolution. He had an obsessive-compulsive personality, avidly collecting and cataloging beetles and other insects around his home. And, his own grandfather, Erasmus Darwin, had already put forth the idea of common ancestry in a poem 60 years earlier:
“The Temple of Nature”
Organic life beneath the shoreless waves
was born and nors’d in ocean’s pearly caves;
First forms minuk, unsceen by spheric glass,
move on the mud, or pierce the watery mass.
These, as successive generations bloom
New powers arquire and larger limbs assume
whence countless groups of vegatation spring.
And breathing realms of fin and feet and wing.-Erasmus Darwin 1802
Most auspiciously, his family was affluent. His father, frustrated with Charles’ unending curiosity and lack of direction, eventually agreed to let him travel abroad with the HMS Beagle even though he considered it to be a complete waste of time. Darwin’s writings of the journey solidified his position as a well respected and well connected member of the scientific community.
He became close friends with Charles Lyell and John Dalton Hooker, the most influential geologist and botanist of his time respectively. Through Lyell, he came to see the natural world as a gradually changing process. Hooker was a trusted confidante with whom he could share some of his troubling revelations. After writing his first essay on natural selection, Darwin told Hooker it was like “confessing a murder”. Hooker gave him the calm, critical feedback Darwin needed to proceed.
The final push came from the naturalist Alfred Russel Wallace who wrote his own version of natural selection in 1858, ahead of Darwin. Both Darwin and Wallace had been inspired by the ideas of the Reverend Thomas Malthus who promoted the idea that population outstrips food supply, and both came to similar conclusions about how this creates selection pressure in nature. Hooker and Lyell arranged to have some of Darwin and Wallace’s ideas presented at a conference in 1858. This inspired Darwin to kick into high gear and finish “The Origin of Species” within a year, unleashing the most consistent and comprehensive explanation of natural diversity ever conceived, then and since.
Darwin did not doubt the literal translation of the bible as a young man, but bravely came to question the nature of the Christian God after observing nature in detail. His attitude was well summarized by his observations of the Ichneumonidae wasp. He wrote:
“I cannot persuade myself that a beneficent and omnipotent God would have designedly created the Ichneumonidae with the express intention of their feeding within the living bodies of Caterpillars, or that a cat should play with mice.”
-Charles Darwin 1860
Like Copernicus’ realization 350 years earlier that the Earth was not at the center of the universe, Darwin’s theory of natural selection pressed heavily against the religious views of his time. Though the religious conflict was never fully resolved in his own mind, Darwin lived to witness his theory gain acceptance in the scientific world and among much of the general public before his death in 1882.
Darwin might have found some solace in the poet Alexander Pope who foresaw the impingement of science upon religion. In “An Essay on Man”, Pope describes Man as passionate and ignorant, existing between God and beast, and that science is a valuable tool for understanding our nature, but not useful in answering religious questions:
Know then thyself, presume not God to scan,
The proper study of mankind is Man.
Placed on this isthmus of a middle state,
A being darkly wise and rudely great:
With too much knowledge for the Sceptic side,
With too much weakness for the Stoic’s pride,
He hangs between, in doubt to act or rest;
In doubt to deem himself a God or Beast;
In doubt his mind or body to prefer;
Born but to die, and reas’ning but to err;
Alike in ignorance, his reason such,
Whether he thinks too little or too much;
Chaos of thought and passion, all confused;
Still by himself abused or disabused;
Created half to rise, and half to fall:
Great lord of all things, yet a prey to all;
Sole judge of truth, in endless error hurl’d;
The glory, jest, and riddle of the world.- Alexander Pope 1734
More:
Tierra
Tom Ray was a tropical biologist who conducted research in the Costa Rican rain forest from 1974 to 1989. His research focused on the ecologies and evolution of various species living there. Eventually he realized he there was a problem with studying evolution in the wild: it occurs far too slowly to actually observe it. He decided therefore to study evolution in a much faster medium, the digital computer. In 1991, he joined forces with the Santa Fe Institute in New Mexico to develop an evolutionary software platform called Tierra.
Genetic Algorithms, programs that simulate evolution to solve a specific problem, had already been well established, but Tierra was different. It wasn’t optimizing anything in particular. Small chunks of machine code were simply left on their own to replicate and compete for access to the CPU, and that was all. Occasional mutations in the copying process allowed evolution to take place. But, this wasn’t a simulation of evolution, these entities were actually evolving. What emerged from Tierra surprised Tom and most of the Santa Fe research team.
The first thing Tom noticed was that these replicating programs became smaller and smaller. A smaller program could replicate faster and so had an advantage over others. Some became so small that they evolved into parasites, tricking other programs into doing the copying for them. The hosts then evolved mechanisms to resist parasites. Some of the host programs were even able to trick the parasites into helping them. Eventually, a form of cooperation emerged where programs helped each other replicate. Then, free-riders emerged who took advantage of this group trust. All of this robust behavior, previously only observed in nature, emerged from Tierra automatically.
Tierra was groundbreaking for the field of Artificial Life, and inspired many systems like it afterwards, including the very robust evolutionary platform, Avida. Most importantly, it gave a demonstration of real evolution occuring in a medium other than nature.
More:
Plastic Brain
How does the brain learn? Understanding that process would allow one to write software that could learn the same way humans do. But, prior to 1949, no one had a very good answer. It had already been well established that the functional unit of the brain was the neuron, and the structure of these neuronal cells had been studied extensively. However, psychologists did not have a good theory about how neurons produced human behavior.
Enter Donald Hebb, a Canadian Psychologist who was fascinated by how the brain worked. Hebb postulated that neurons form cell assemblies, collections of neurons that act in concert to produce behavior. This idea formed the beginnings of the field of connectionism, an approach to the mind that views complex behavior as emerging from an interconnected network of simpler units. But how do these networks form? To answer this, Hebb proposed a mechanism which has come to be known as “Hebbian Learning.” The idea stated simply is: “Neurons that fire together, wire together.”
The brain is fully connected at birth, but the strength of these connections changes through time as we learn, forming the cell assemblies that Hebb theorized were responsible for behavior. Hebbian learning postulates that when neuron A activates, and then causes neuron B to activate, then the connection strength between the two neurons is increased, and it will be easier for A to activate B in the future. The idea sounds simple, but it goes a long way in explaining how neural networks form in the brain. Not every learning process in the brain can be explained by Hebbian learning, but it does provide an explanation of how complex networks of neurons could form.
After Hebbian learning made its debut in the 1949 book, “The Organization of Behavior,” it then became possible to program computers with the Hebbian rule, giving them the ability to learn. Today, many different types of artificial neural networks (ANNs) are used extensively in the field of artificial intelligence, including applications in face identification, speech and handwriting recognition, financial applications, data mining, and even autonomous vehicles. Hebb’s discovery spawned a whole branch of artificial intelligence and methods for constructing learning mechanisms on computers. ANNs are not yet sufficient for creating human-level intelligence on a computer though. Real neurons are complex biochemical engines, whose behavior can only be approximated with ANNs. Also, human brains come pre-configured to some degree, and without understanding this innate structure well, building large artificial neural networks is not practical.
More:
Simple Life
In 1986, the German neuroscientist, Valentino Braitenberg, made an important observation about human psychology. He said, “When we analyze a mechanism, we tend to overestimate its complexity.”
Try following an ant around while it searches for food. You might infer a complex mind that is processing a myriad of difficult decisions. However, the ant’s actions can be explained by a very simple set of rules.
Even so, it’s not always easy to determine what those rules might be, looking in from the outside. Braitenberg therefore concluded that, when studying complex behavior, you must invent your own rules, from the inside looking out. In short, you tinker from the simplest possible starting point and see what emerges.
This line of thought led to his famous book, “Vehicles: Experiments in Synthetic Psychology,” which details a series of thought experiments in which simple mechanisms seemingly exhibit complex behavior.
Consider, for example, Braitenberg’s vehicle #2a. It’s a two-wheeled robot with light sensors. When light on the right side increases, so does the right wheel’s speed. Same with the left side. The result is a robot that avoids light. Cross the wires, and you have a robot that steers and accelerates towards a light source (vehicle #2b). While the simplest of parts comprise both robots, Braitenberg describes the former as “cowardly,” and the latter as “aggressive.”
How about robots that display “love?” Braitenberg’s vehicle #3 adds some inhibitory connections, which cause the robot to approach a light, and then gradually stop to stare at it indefinitely.
Other adjustments can create robots, which Braitenberg describes as “exploring, knowing, displaying instincts, making decisions, having foresight, ego and optimism.” Obviously, these are subjective, anthropomorphic terms, but Braitenberg purposely uses them to make his point: complex behavior can arise from simple mechanisms, IF they are configured just the right way.
More:
Minimax, Theseus, and the Juggler
Claude Shannon (1916-2001) was one of the greatest engineers of the 20th century. He is considered the “Father of the Information Age,” inventing the concepts necessary for processing information digitally. Prior to 1948, no one had seriously considered sending or storing information digitally. It took decades for technology to catch up fully with his ideas. Had Shannon never existed, we would still have cell phones, MP3′s, hard drives, high speed communications, and the internet, but it’s likely that these technologies would have been delayed by many years.
Like many geniuses, Shannon had many strange hobbies and interests. He built a gasoline-powered pogo stick, for example. He also built various types of computers: one that could process Roman numerals; and another that tried to solve the Rubik’s Cube. He co-invented the wearable computer, and also had an interest in gambling. He used to make weekend forays to Las Vegas with Ed Thorp, the mathematician who proved it was possible to beat the house in Blackjack through card-counting. Shannon also had a side interest in Artificial Intelligence and made several significant steps in the field. Highlighted here are Shannon’s most important contributions to the field of AI:
Shannon’s Chess Algorithm
In a 1950 paper, Shannon worked out the first complete recipe for programming a computer to play chess. In his paper he described the so-called minimax algorithm, which looks into the future for possible counter moves and always assumes the opponent will make his best possible move. It remains the de facto method used by chess programs, and was even used by Deep Blue, the chess machine that eventually defeated the world champion, Gary Kasparov, in 1997.
Shannon’s Mouse
Fascinated with the idea of machines that could learn, Shannon built a mechanical mouse, dubbed “Theseus,” for the legendary king of Athens who escaped the labyrinth of the Minotaur. It was the first learning device of its kind. On its first run through the maze, the mouse would use trial-and-error to find the target. The second time, however, the mouse would find the target faster by using its prior experience.
Shannon’s Juggler
Shannon became somewhat obsessed with juggling in the 1970s and conducted the first serious mathematical analysis of the subject, which culminated in his “Juggling Theorem.” To test his theorem, he built a robot in the likeness of WC Fields, which could bounce-juggle three balls automatically. More of an exercise in dynamical systems than artificial intelligence per se, the device had no feedback mechanisms, but was able to maintain a stable juggling system indefinitely.
Claude Shannon was yet another brilliant mind, who helped transform the world we know today. He lost his battle with Alzheimer’s in 2001.
More:
Talk to Me, ELIZA
How do you know if a computer has achieved human-level intelligence? Alan Turing, the founding father of computer science, came up with a simple method: open two chat windows, one with the computer and another with a human being, and try to distinguish which one is which. If you can’t tell, then the computer has passed the test and can be called “intelligent.”
Writing a program that can pass the Turing test has been the holy grail of Artificial Intelligence (AI) since 1950. There’s even an annual competition, where researchers can submit their AI programs to see if they pass the test. None ever do, but some get close, and awards are given to the best ones.
Turing was a genius and way ahead of his time, but The Turing Test suffers from one major flaw: passing the test is more about fooling people than achieving true AI.
For example, in one chat window you might ask, “What is 68185 divided by 13?” If it’s the computer you’re talking to, it might purposely take a long time to respond, or send the wrong answer back, or even reply with, “Gee, I failed math in third grade.” This is all just trickery in order to get you to think the computer is actually human.
Similarly, if you were to ask the computer an abstract question like, “How do you feel when you hear classical music?”, it could search for certain keywords like “how do you feel” and then pick from a set of canned generic responses, to respond with, “I feel wonderful,” which simply creates the illusion that it fully understands the question.
ELIZA was a program written by Joseph Weizenbaum in 1966, which was exceptional at pretending to be human. Weizenbaum’s idea was to simulate a therapist, who encouraged the patient to do most of the talking. It worked by scanning for certain keywords typed by the user, and then giving either a generic response or repeating an earlier comment, to give the impression that it was listening.
ELIZA worked so well, that people formed emotional bonds to it. His test subjects took offense when he asked to review the transcripts, saying it was an invasion of their privacy. Some asked him to leave the room while they shared their innermost feelings with ELIZA. Others called back later saying the therapy had helped them and begged for more sessions.
The Turing test was a significant step in the field of AI. Its style, which assumes that intelligence is nothing more than symbol manipulation, or following a set of preset rules, is now referred to as Good Old Fashioned Artificial Intelligence (GOFAI). However, the idea that disembodied programs divorced from their environment can achieve human-level intelligence is no longer in favor with AI researchers today.
More:
Robots That Balance
Have you ever tried to balance a broom on one finger? Humans can do this fairly easily, and so can machines because the control system equations for this “inverted pendulum” system are relatively simple. It turns out that the process of walking and running in animals and robots is very similar to balancing a broom. If you assume the body of the robot is always about to fall over, then the inverted pendulum equations can be used to compute where the legs should be at each moment in order to maintain balance.
Marc Raibert, who founded the MIT Leg Laboratory in the 1980s, used this idea to build a robot with a single leg that could maintain its balance. Although it looked like a hyperactive pogo stick, the “hopper” robot was a success. Expanding on this work, he later moved to two legs, and finally four:
In 2005, Raibert founded Boston Dynamics, makers of the Big Dog military pack robot which uses the same basic concepts for remaining dynamically balanced. In fact, it can stay balanced on slippery surfaces and even remain stable after receiving a sharp kick. Boston Dynamics is funded by DARPA. The pack bots are designed to assist soldiers by following them in the field and can carry up to 120 pounds of supplies across a variety of terrains.
More:
Elephants Don’t Play Chess
Artificial intelligence research was in a slump in the 1980s. The most popular design paradigm for intelligent robots had been developed decades earlier with machines like Shakey and the Stanford Cart. Hulking boxes on wheels, these machines relied on expensive sensors and computers that would take hours to navigate a room. To move just a few feet, the robots had to scan the room and perform billions of calculations. AI researchers resorted to making incremental improvements on this basic design, hoping that faster computers and better programming would eventually make them more practical.
Rodney Brooks, a researcher at the MIT Artificial Intelligence lab in the late 80s, knew something was fundamentally wrong with this approach. Human beings take seconds, not hours, to walk across a room, and yet our brains are slower than a computer’s.
In a revolutionary paper titled “Elephants Don’t Play Chess”, Brooks turned the field of robotics on its head by introducing the idea of behavior based robotics. In this bottom-up design approach, the robot interacts directly with the environment, instead of scanning it and performing numerous complex calculations. In place of a single large brain controlling everything, his robots linked together several tiny brains, each sensing and reacting to the environment in real-time.
One of Brooks’ prototypes was Genghis, a six-legged robot that could walk, explore, and climb over obstacles using inexpensive hardware. Through Genghis, he demonstrated the feasibility of his design methodology, which he called the subsumption architecture. Many AI researchers at the time were not impressed and dismissed his work as uninteresting.
Brooks later co-founded iRobot, the makers of the Roomba robot vaccuum cleaner. Using a behavior based design, the Roomba has no information about the room it is vacuuming. Instead, it reacts to the environment as it moves, altering its behavior slightly when it bumps into chairs or reaches the edge of a stairwell. The bottom-up design reduced manufacturing costs enough so the robot could be priced for the mass market. With millions of units sold, the Roomba was the first truly successful consumer robot product. Brooks remained the CTO at iRobot until September 2008, when he left to found his newest company, Heartland Robotics, in Cambridge, Massachusetts.
More:
Passive Dynamic Walkers
In the late 80s, Tad McGeer invented the concept of passive dynamics. While doing graduate work at Simon Fraser University, he built a robot that could walk down a plank without power, sensors, or a control system. The robot was built from metal rods, springs, and weights in just the right configuration such that the legs and arms would swing in a coordinated way as it ambled down. It was also able to walk efficiently on a flat surface by giving it a small push. In 2001, Steve Collins from Cornell University build a passive walker based on McGeer’s original model shown in the video below.
The idea of adding power to passive dynamic walkers later inspired other universities to develop their own versions. McGeer demonstrated that a bottom-up approach to robotics must begin with the body. A cleverly designed body can remove the need for high speed computers to control all the leg movements. This approach has still not been widely accepted. For instance, Honda’s Asimo Robot which after 20 years of research and at a cost of $1 million per unit still uses a bulldozer approach of scanning and conquering the environment instead of dynamically interacting with it.
More:
Grey Walter’s Tortoises
Dr. Grey Walter was a neurologist, robotics pioneer, and a bit of a mad scientist. Living in Bristol, England in 1949, without the aid of modern day computer processors, he built reactive, autonomous robots that could wander about and avoid obstacles. Each robot had two simulated neurons, sufficient for them to display complex behavior. Significantly, Walter’s tortoises represent the first real world demonstration of artificial life.
More:
