Everyone is familiar with the summer of ‘69, but can you recollect the historical event that took place in the summer of ‘56? Artificial intelligence might’ve been a term you only came to know in the 21st century, but this research field was actually founded at a summer workshop at Dartmouth College campus. During this workshop, attendees predicted that a machine matching human intelligence would be invented in the next generation. Although this workshop wasn’t as successful as hoped, it planted the seed for future AI research and blazed the trail towards humanity's fourth industrial revolution.
Artificial intelligence is a branch of computer science that refers to programmed machines with the ability to simulate human intelligence or exhibit traits normally associated with a human mind. AI is a term that encompasses a broad spectrum of technologies; for example, machine learning, neural networks and computer vision, to name a few. Thus in order to better understand the concept of AI, one must look at the various stages of intelligence as well as types of artificial intelligence.
There are three main stages through which AI learns and evolves: narrow intelligence, general intelligence and super intelligence.
Narrow AI is a term used to define AI systems that are programmed to handle a singular task. American philosopher John Searle describes narrow AI as being “useful for testing hypotheses about minds, but would not actually be minds”. Narrow AI is common in the world around us, and can even be found on your iPhone. Siri is an excellent example of narrow AI, as the algorithm integrates machine learning on a mobile platform but is unable to complete tasks outside of its preprogrammed set of abilities.
General AI is a stage of artificial intelligence that includes systems with adaptable intellect that are able to carry out a myriad of tasks. This stage of AI is level with human intelligence and has yet to be invented. The idea of AI general intelligence has sparked massive controversy in the scientific community, fueling debates on whether it should be embraced or feared. Some have even questioned whether its emergence might spell the end of the human race. Stephan Hawking roughly stated that the development of full artificial intelligence will lead to AI “taking off on its own, and redesigning itself at an ever-increasing rate. Humans, who are limited by slow biological evolution, couldn’t compete and would be superseded.”
Although we don’t quite share Hawking’s fears regarding general intelligence, AI super intelligence might just spur us into prepping for a potential Skynet attack in the near future. Super intelligence is a proposed form of artificial intelligence that encompasses systems with cognitive abilities exceeding that of the human race. Although intelligence shouldn’t be feared, one can’t help but wonder what the future might hold once humanity is superseded as the most intelligent lifeform on earth.
Due to AI encompassing a wide variety of technologies, the most logical way to classify AI types is based on functionality. The four widely accepted types of artificial intelligence are: reactive machines, limited memory, theory of mind, and self-aware AI.
Reactive machines are the most basic type of AI system, in that they cannot function beyond the task they were programmed to perform and can only react to pre-programmed situations. An example of reactive machine AI is IBM’s DeepBlue, a chess-playing supercomputer. Created in the 1980s, DeepBlue was able to explore over 200 million possible chess positions per second and later became the first computer to defeat a human opponent, beating Russian chess grandmaster Garry Kasparov in 1997.
Limited memory AI systems are mostly machine learning models that operate on a combination of observed data and pre-programmed data. Autonomous vehicles are an example of limited memory AI, in that they learn to drive and function among other vehicles and are able to adjust as necessary to external factors; for example, slowing down when pedestrians are crossing the road or reacting to traffic signals.
Theory of mind AI systems exhibit human-level decision making abilities. Currently, artificial intelligence can only exhibit humanlike capabilities, and thus this type of AI has not yet been fully developed. Sophia the Robot, created by Hanson Robotics, is an example of AI pushing humanity closer towards theory of mind. Not only does Sophia almost look human, but she is also able to recognise images and respond to interactions with the appropriate facial expressions, thus exhibiting a level of emotive ability.
Self-aware systems are the most advanced form of artificial intelligence – they take theory of mInd AI a step further. This proposed form of AI has yet to be developed, but it would encompass AI systems with human-level consciousness and the ability to exhibit human-level comprehension and replication. It would also be able to think and react independently. Many believe that self-aware AI is the ultimate goal of AI development, but the implications of a machine with self-awareness are uncertain. Not only does self-aware AI raise concerns regarding the safety of the human race, but ethics also come into play. After all, if a machine can think and feel as humans do, essentially being conscious and relating to other lifeforms as equal, it would become unethical for another human being to “own” it.
Who would’ve expected that a small workshop in the summer of ‘56 would push humanity into a new era of tech evolution. Artificial intelligence has already seeped into modern society and taken root in numerous corporations, but when taking into account the stages and types of AI we have yet to reach, it’s clear that humanity has only dipped it’s toes in the pool of possibilities AI advancements will offer. What are your thoughts and opinions on artificial intelligence? Do you feel this generation will see the emergence of super intelligence and self-aware systems? All we can do at this point in time is speculate, but who knows, maybe future generations will cite the summer of 2020.