As humans, we delight in the idea of our uniqueness, our exclusivity, our singular special as-yet undefined trait which enables us to dominate the bejesus out of this planet. In decided contrast, the science of Artificial Intelligence
argues we may not be as inimitable as we hope. 1956’s Dartmouth Summer Research Project
on Artificial Intelligence proposed its field be founded on the claim that humanity’s central property, it’s vaunted intelligence, can be described with such precision that an automation could simulated it without flaw.
To those so much as transiently influenced by popular culture, the mention of AI will trigger the automated response “Oh, like Skynet
?!” Indeed, James Cameron’s
fictitious military defence system began learning exponentially when brought online, acquiring self awareness in only 25 days.
There’s scientific credence behind this conceit, with Demetriou’s Theory
stipulating Self Awareness is a major factor in the development of inferential processing. Thankfully, the overwhelming majority of creatures with self awareness (Humans, Dolphins, Elephants, Magpies) don’t celebrate this by bombing the Russians... To the best of my knowledge.
Over the preceding 56 years, the on-again/off-again love affair assorted Government funds enjoyed with AI’s technological advancements none-the-less yielded some significant tools... Ever heard of this indie Tech company called GOOGLE
The wider notion of Search and Optimization
was pioneered via AI research. Logic
programming and automated reasoning came about by trying to make computers more ‘human’ while Artificial Neural Networks
(interconnecting neurons and nodes) directly mimic those brain cells we’re so fond of drowning. Finally those of you reading this very feature on the wonder of PC/tablet/mobile can thank AI for developing computer programming languages such as Prolog
More recently (if not entertainingly) Xbox’s Kinect
employs its 3D body motion interface by way of algorithms which emerged after extensive AI research. NO SPHERE OF INDUSTRY IS SAFE FROM THE AI MENACE!
But there’s no IMMEDIATE
need to take to the streets, armed with a sharpened broom handle and a sock full of pennies, intent on mutilating any motherboard to cross your path. The happy truth is GLaDOS
remains a fictitious, if hilarious, invention. A Genetic Lifeform and Disk Operating System, built for the testing, maintenance and eventual murder of Aperture Science
iconic AI echoes 2001
’s markedly more sympathetic HAL 9000.
Back in 1968 when Kubrick
fantasised about Discovery One’s
omnipotent AI capable of singlehandedly sustaining an orbital station while communicating flawlessly with its crew, his prediction was met with acceptance, not ridicule. AI Pioneer Herbert A. Simon
postulated the globe would see Artificial Intelligence
‘by 1990 doing any work a man can do”, notably natural language processing, lip reading and advanced reasoning.HAL
was also a dab hand at chess. In this sole regard, Kubrick’s
vision was prophetic.
Modern day AIs cannot match GLaDOS
for wit or charm. Sorry Siri
! Yet they have been applied, with admittedly varying success to processing and recognition issues...OCR
: Optical character recognition is basically how a computer reads and converts pen scribbled words into glorious digital data. Typewritten text is still not 100% accurate, even where clear imaging is available however commercial OCR
software typically varies between 71% and 98% accuracy. The computers. They are catching us...Natural Language Processing
: Even a four figure IQ is irrelevant without a reliable mode of communication. Concerning interactions between Computers and Human Language, modern NLP algorithms are based on statistical machine learning. Automatically learning language rules by analysing sets of documents, hand-annotated with the correct values, NLP has significantly simplified the contemporary interface process.Game Theory
: If Roger B. Myerson
is to be believed, (and for the laugh, indulge him) this is "the study of mathematical models of conflict and cooperation between intelligent rational decision-makers." Numerous logical theories have a basis in game semantics, while computer scientists have used games to model interactive computations. Also, game theory provides a decent theoretical basis to the field of multi-agent systems -multiple interacting intelligent agents (AIs) within an environment. In layman’s terms: QUAKE
: When AI’s develop algorithms and techniques that are able to determine whether the behaviour of a system is correct, they’re going House
on its ass! While diagnostic AI’s probably shouldn’t be trusted to run Nuclear Plants, they’re frequently consulted in medical fields and aint none too shabby at scanning your lappy for viruses!
In Science fiction, the ultimate AI likely resides in the Culture Novels
of Iain M. Banks
. ‘Minds’ are afforded the same rights and respect as similarly self aware creatures, drones, humans etc. However the standard Mind of a Culture General Systems Vehicle
houses roughly 1 million yottabytes of information.
Perspective time...Where a gigabyte is 10 to the 9 and a terabyte is 10 to the 12, an Exabyte is 10 to the 18. As of 2009 the Internet’s entirety was said to house 500 exabytes of information. A SINGLE yottabyte is actually 10 to the 24.
A more outlandish fiction is hard found, but I intend to conclude on the chilling subject of Artificial Intelligence contesting Human counterparts. For comparison, Subject Matter Expert Turing Tests
have evaluated AI performance as followed.
Regarding Driving, Poker, Translation and speech recognition, AI lags behind. It’s probably the clutch what’s throwing them... However in the spheres of Scrabble, Jigsaws, Quizes and even Chess, no human yet lives to match our AIs.
And perhaps most alarmingly, Artificial Intelligence
has received the rank of optimal (impossible to improve on its performance) on the following tests: Checkers, The Rubic’s Cube and, wait for it, Connect 4.
This is proof. Judgement Day is coming. And it’s riding a diagonal wave of quadrupled, plastic red and yellow discs!