Artificial Intelligence…past and future
By Paul Field – CEO @ Improved Apps
We may think artificial intelligence is a new concept for the computer age, but actually, the notion of AI is one that has fascinated mankind for centuries. It has managed to achieve the dubious status of being the stuff of both science-fiction but also horror stories.
As far back as classical Greece there were stories of something man-made (that is ‘artificial’) being created with intelligence. Crete was supposedly protected from pirates and invaders by a giant bronze automaton called Talos. At the same time. The Greeks were making computers such as the Antikythera mechanism that were able to accurately predict the position of heavenly bodies and eclipses. In pharaonic Egypt, mechanical statues made for temples were thought to be possessed of wisdom and emotion.
Fast forward to the 12th century and the Islamic scholar Al-Jazari (author of the ‘Book of Knowledge of Ingenious Mechanical Devices’) was creating sophisticated automata that could mimic human movement and even created a mechanical orchestra. In the late sixteenth century in Prague, Rabbi Judah Loew was said to have created the golem: A creature of clay that was animated by written instructions placed into its mouth.
The mechanical pace then hots up with Blaise Pacal’s mechanical calculator in 1642 and Leibniz producing the first mechanical reckoner (since the Greeks?) in 1672. From then to now, we pass quickly through an exploration of the morals of creating sentient beings in Frankenstein, the creation of the first computer in the 1850s by Charles Babbage, the first use of the word ’robot’ in 1920 by Karel Capek and then onto the father of modern thinking on AI: Turing – he’s famous now for his work on cracking the enigma code used by Germany in WW2, and addressed the problem of artificial intelligence, and proposed an experiment that became known as the Turing test, an attempt to define a standard for a machine to be called “intelligent”. The idea was that a computer could be said to “think” if a human interrogator could not tell it apart, through conversation, from a human being. Turing even wrote a chess-playing program in advance of their being a computer capable of running it.
The exposure we have to AIs comes from many different sources both fictional and real. A quick search will highlight AI in many diverse references: Films such as 2001 A Space Odyssey with HAL, (ask Cortana if she is related to Hal and she will ‘politely dodge the question’), Cortana herself (the AI from the video game Halo – borrowed by Microsoft), Westworld with its vision of a ‘slave uprising’, Deep Blue – the all-conquering chess playing computer, SkyNet (the AI that tried to take over the world in the Terminator series) and up to date with Watson the game-playing AI.
It is only really with Turing and beyond that there has been a real attempt to define what intelligence is and what it might mean for a created thing. We can discount the mechanical machines as nothing more that pre-programmed devices, disregard the limited function, human analogue robots and end up with what? The answer is somewhere between the single purpose deliberation of the golem and the all-encompassing ambition of Skynet.
Our yardstick for artificial intelligence is if course ourselves. If we look at the performance of the human brain compared to computers we get a very interesting picture – (see above – courtesy of Scientific America): Our brains are actually extremely power-efficient, but can no longer compete with the sheer processing speed of today’s largest super computers (but can beat your desktop by some distance). It is however, quite surprising how recently we have been overtaken and by what are still relatively small multiples. (Even a cat’s brain beats the newest iPad—1,000 times more data storage and a million times quicker to act on it).
However, this rather misses the point. Human brains are capable of performing a vast number of tasks, whereas AIs are still largely one-dimensional. The connectedness of the different parts of the brain mean that we can still out-think computers unless the context is heavily rule-reliant and all permutations and variables are known, and it becomes simply a matter of number-crunching (such as working on medical diagnosis or stock markets). AIs have started to become ‘creative’ but this is not really the creativity or original thought of a Mozart, but more the ‘song-smithing’ of the elevator music composer.
It is clear that AIs have the capability to provide a great service to us in the realms of analysis, interpretation, prediction and automation of certain processes. As AIs now realistically have the chance to outstrip humans by increasing distances, we need to be sure that we have identified the right areas to apply the AI and the right controls to prevent the runaway horror vision of fiction.
Paul Field – CEO @ Improved Apps