Artificial Intelligence has been around since the 1950's. Alan Turing envisioned a machine that could think. He devised a test, aptly named the Turing Test, published in an article titled Computing Machinery and Intelligence. He proposed the notion that a computational machine could answer a series of questions from a panel of judges. The responses would be rational, thoughtful, and indistinguishable to another human. Prior to that, Turing spent a decade creating the blueprint for "machine intelligence".
John McCarthy, a distinguished professor at MIT coined the term Artificial Intelligence and organized an international conference dedicated to the pursuit of AI. There he met Marvin Minsky and together they worked to advance the theories and concepts of bringing AI to life. They used the LISP language as their programming language of choice.
They ran into some obstacles, included limited processing power, limited storage capacity, high costs, lack of funding and the underlying complexity surrounding the concepts involved.
In the mid-1960s, an MIT professor, Joseph Weizenbaum, created a computer program named ELIZA. It simulated a virtual "doctor" and was able to interpret natural language input with a somewhat intelligent set of responses. Although limited in functionality, it gets credited for being one of the first AI programs in existence.
In the mid 1980's, a concept known as Backpropagation was created. This technology leveraged complex algorithms to process information based on a known set of data. The program received a set of input data, flowed through a series of Neurons, which performed a calculation, produced a number between 0 and 1, and based on the configuration, it either fired a signal to another connected neuron, similar to the synapses in the human brain. It then flowed into a series of "hidden" neurons which also had complex calculations. Finally, the data flowed to the end and produced a final response. That response was compared to the known data and if differences appeared, the set of data was flowed backwards through the neural network to recalculate the weights on each of the neurons. Soon, multi layered neural networks were created to increase the capacity and accuracy.
In 1984, the entire field of Artificial Intelligence slowed down. At a conference of American Association of Artificial Intelligence, the term AI Winter was used to define this stagnation in the field. Basically, the hype surrounding AI was under scrutiny by the funding groups such as government bureaucrats and venture capitalists. This pessimism pushed the AI field into obscurity for some time.
In the early to mid-1990s, AI was becoming known in the business world. Two of the main reasons were: the increase in compute power; and isolating specific problems within specific domains.
“An Intelligent Agent is a system that perceives its environment and takes action which maximize its chances of success.”
In 1997, IBM's Deep Blue knocked off the world chess champion Garry Kasparov.
Strong vs. Weak Artificial Intelligence
In the pursuit of Artificial Intelligence, there are basically two camps.
Weak or Narrow Artificial Intelligence
Personal Assistants typically reside on computers and smart phones. They can learn the behavior of the individual using the application. Preferences, places of travel, history of searches and browser trails, the application is able to inform its users to alter a course of action based on given parameters in real time. These assistants are becoming more precise as they are embedded into everyday applications.
Strong or Artificial General Intelligence
The second type of AI is known as Strong AI or Artificial General Intelligence. Strong AI or AGI is the intelligence of a machine that match or surpass the abilities of a human being in performing tasks. Some of its characteristics are having the ability to learn, reason, communicate in natural language, possess creativity, have morals and be self-aware.
When computers become aware of themselves, they will be able to recursively build machines that are more intelligent than themselves. This will lead to an exponential increase in the pace of progress, eventually moving the intelligence of machines beyond the comprehension of human beings. This hypothetical event is commonly referred to as the Singularity.
Artificial Neural Networks
Artificial neural networks (ANN) are about function approximation. Basically, you have a Neural Network. The Neural Network takes in Input. That input is interpreted by Neurons. These neurons have approximated weights. Based on the results of the calculation, if they exceed a specified threshold, it fires a 1 or 0 as output, which is sent downstream to other Neurons. It's possible to over-train a model, in which the output has to contort itself drastically, which sends the results into a tailspin and results become nonsense.
Due to the complexity of the human brain, researchers have not been able to reproduce it at this point in time. Artificial Neural Networks attempt to simulate that complexity, with some level of accuracy. There are pre-canned packages you can purchase which do all the heavy lifting for you, exposing this technology to more people.
Artificial Quantum Neural Networks
Researchers are combining Quantum Physics with computers. Although this research is in the early stages, these computers are known as Quantum Turing Machines. In classical computing, everything is based off of the concept of “binary” or being in the state 0 or 1. In the Quantum world, the binary unit is replaced with a unit named Quantum Bit or Qubit for short. When Quantum Mechanic principles are applied, the neural unit can result in more than two states, 0, 1 or both. This has great implications as these new systems can perform calculations extremely fast and can actually solve some problems deemed impossible in the classical binary approach. One example is the ability for quantum computers to decrypt public keys, which are the foundation for internet security today. The underlying rules that make up Quantum physics are quite complex. One company leading the charge with Quantum computers is D-Wave. They define Quantum Computation as follows:
“Rather than store information as 0s or 1s as conventional computers do, a quantum computer uses qubits – which can be a 1 or a 0 or both at the same time. This “quantum superposition”, along with the quantum effects of entanglement and quantum tunneling, enable quantum computers to consider and manipulate all combinations of bits simultaneously, making quantum computation powerful and fast.“
This cutting edge technology is making strides in problem solving and could potentially be used to advance the world of Artificial Intelligence by leaps and bounds.
Morals and Ethics
With the rise of intelilgent machines, some thought needs to be spent on the Ethical consequences. How will AI and Robots behave amongst humans? What will determine the moral blueprints for acceptable behavior? What if an AI kills a human? Would it go to machine prison? Could it get married? Or buy insurance? Who owns it? Will machines have funerals? Can a machine be sold? What if it steals? Should they be entitled to vote? What if your machine gets stolen or abducted? Can they reproduce? If so, are they responsible for the care of their youth until they graduate from High School? Can you euthanize a robot? Do robots get paid for services rendered?
Every day we get closer to the reality of Artificial General Intelligence. At some point in the future, machines will be integrated into our society. It's up to us, now, to determine the roles, rights, duties and responsibilities assigned to our new intelligent beings.
Current AI Organizations
One organization dedicated to the pursuit of Artificial Intelligence was created by Paul Allen, one of the original founders of Microsoft, called Allen Institute for Artificial Intelligence with the moto:
“Our mission is to contribute to humanity through high-impact AI research and engineering.
“OpenAI is a non-profit artificial intelligence research company. Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return.”
“We’re committed to advancing the field of machine intelligence and developing technologies that give people better ways to communicate. In the long term, we seek to understand intelligence and make intelligent machines.”
Microsoft Research has an Artificial Intelligence (AI) Group.
“The Artificial Intelligence (AI) group consists of an elite team of researchers who have strong expertise in artificial intelligence, machine learning, game theory, and information retrieval. The group is devoted to the following research directions: large-scale distributed machine Learning, cloud computing, robot, game-theoretic machine learning, and deep learning techniques for text mining.”
IBM has a research team, going back to the 1950’s when AI was first introduced. IBM is known for their Cognitive Machine called Watson. Another leading tech company, Baidu, has a research facility in Silicon Valley called Silicon Valley AI Lab.
Intelligent Machines have grown since the early days of the mid 1950's. With the increases in storage capacity, compute power, accessibe software, shared knowledge and technogology advances over the past 60 years, we've witness the rise of Artificial Intelligence into smart applications called Personal Assistants. How soon until we make another leap into the world of Artificial General Intelligence, where machines can learn and interact in real time and pass the Turning test? How soon will machines work side by side with humans to solve complex problems, reduce costs and make the world a better place. True artificial intelligence could be integrated into mainstream society sooner than we think.