Artificial Intelligence - has its time come?
(Notes and links used in a talk given at the Javea Computer Club - April 12th 2012
2
Intelligence: A definition
Capacity for learning, reasoning, understanding, and similar forms of mental activity; aptitude in grasping truths, relationships, facts, meanings, etc.
Artificial intelligence
= man made machines which learn, reason ….and "understand"
Are we really in a position to create such things?
3.
Intelligent mechanical beings in legends of ancient times - seem to have been an obsession (still are)
http://en.wikipedia.org/wiki/History_of_robots
Chinese accounts relate a history of automata back to the 10th century BC when Yan Shi is credited with making an automaton resembling a human in an account from the Lie Zi text.
Mechanical servants built by the Greek god Hephaestus (Vulcan to the Romans), the clay golems of Jewish legend and clay giants of Norse legend.
Aristotle speculated that automatons could someday bring about human equality by making the abolition of slavery possible in his book Politics (ca. 322 BC).
4.
In science fiction
It is believed that one day we will be able to make Robots or Androids (human shaped robots) which think like we do.
Asimov: Three laws of robotics
1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.
Asimov's Robot detective
http://en.wikipedia.org/wiki/R._Daneel_Olivaw
These robots are "perfect citizens"
5
Science Fact
Until recently - we didn't have the technology to even dream about building an intelligent machine. (1960's computer memory was in kilobytes and processing speeds slow); also we knew very little about how the brain was put together and how it worked.
We are just beginning to understand now… But the human brain is probably the most complex thing known to Man - do we have any chance of building something like it ?
In 1950 Turing: posed the Question "Can machines think? "
Actually this got modified a bit into the Turing Test:
http://en.wikipedia.org/wiki/Turing_test
A person has a typed conversation with a human and a machine. If he can't tell that the machine is a machine, then the test will have been passed.
Basically can a machine fool you into thinking that it is human?
There is a $20,000 prize for a bet:
“By 2029 no computer - or "machine intelligence" - will have passed the Turing Test.”
Bet laid in 2002 (27 years to make that machine (by which time 20000 dollars won't be worth much!)
6.
But according to John Searle's Chinese Room thought experiment. The test could be passed - but no intelligence would be required.
The experiment is the centerpiece of Searle's Chinese room argument which holds that a program cannot give a computer a "mind", "understanding" or "consciousness", regardless of how intelligently it may make it behave.
We see this type of computer "Intelligence" in Google Translate etc. … and in students taking multiple choice exams…
So computers can behave as if they have "minds" - whatever that is. Can they interact with human beings in a day to day environment ?
7.
There have been some pretty inane attempts at making interactive robots
“Alice” The Artificial Linguistic Internet Computer Entity:
Touchy feelie with a robot:
Honda's Asimo Robot The best of the lot.
Can move very well
…. and it is now autonomous (previously controlled by an operator)
The second clip may not be very impressive, but demonstrates how difficult "autonomy" is.
8.
You'd have trouble dancing cheek to cheek with ASIMO or giving it a cuddle.
So, some researchers are trying to make robots that move the way that people do, so that they can interact more naturally with us.
e.g
ECCEROBOT (Embodied Cognition in a Compliantly Engineered Robot) (!!!)
the first "Anthropomimetic" robot (…looks a lot like Terminator)
Others are developing robots that can learn.
e.g. Luc Steels (See BBC 2 documentary on Artificial Intelligence with Prof Sautoy)
9
Can computers ever be smart enough ?
There have been several high profile "experiments" pitting man aginst computer.
Chess :
IBM's “Deep Blue” Super Computer vs Gary Kasparov (who had never been beaten)
1997 – Kasparov 1 – Deep Blue 2, Draw 3
http://www.research.ibm.com/deepblue/
Quiz Game:
Watson Computer wins at a quiz game - "Jeopardy"
Note: Watson always has a nice, stable, disposition.
There are possible uses of this technology in Law, medicine, Engineering etc.
Human judges tend to give harsher sentences before lunch than after lunch.
They can also be primed by irrelevant factors, such as whether they have seen high scores on a pair of dice, (harsher sentence) not low.
(See PDF paper on Internet: Playing Dice With Criminal Sentences: The Influence of Irrelevant Anchors on Experts’ Judicial Decision Making )
Scary heh ? - If we used the "Watson" technology, maybe we can get fairer and more reliable sentencing and better medical diagnoses.
10
Current computing power in relation to brain power is still low
Biggest computer Fujitsu K Computer:
will have 800 computer racks when finished. Equivalent of linking 1 million PCs = 8.2 petaflops and will use enough energy to power 10,000 houses. Huge thing.
Human brain:
Guesstimate: 100 - 1000 petaflops
(1 Petaflop = 8 quadrillion calculations per second….
8000,000,000,000,000)
Power consumption 20-40 watts
Size - as big as yer head.
If you combined all the supercomputers in the world you get 58.9 petaflops - or half a human brain.
11
Storage
Storage - we still don't know how human memory works - It is different from digital storage and based on connections / synapses between nerve cells and patterns in the ways they fire. The basic structure determined by our genes.
Digital storage remembers everything.
Human memory - selective, we can forget and sometimes remember things that never happened.
12
Modelling Brain function using computers:
So that we can better understand our brains.
Blue Brain Project:
http://bluebrain.epfl.ch/
Based around an IBM supercomputer - but, as they say - Each simulated neuron requires the equivalent of a laptop computer. A model of the whole brain would have billions of neurons.
Spain involved in this with the Cajal Blue Brain project
http://cajalbbp.cesvima.upm.es/
Providing information at the cellular and molecular level.
C2 Cortical simulator:
This has succeeded in making a model "smarter than a cat"
Problem with these models is that when they get close to simulating the real thing - they get as complicated as the thing they are modelling:
It just means that instead of having one thing you don't understand you now have two things you don't understand: the real system, and a simulation of the system that has all of the complexities of the original plus an additional layer of complexity associated with the model's implementation in hardware and software.
This is called Progress!
13
Human beings are not just brains - but brain + body - senses / hormones / coordination / emotions / learning/
Perhaps we should get away from thinking about the human brain, and trying to create artificial intelligence which mimics it.
Why not make a different kind of intelligence ?
e.g.
Driverless Trains:
All over the world - especially in underground systems, including Valencia.
Driverless bus:
http://www.sciencemuseum.org.uk/antenna/driverlessbus/111.asp
Driverless Cars:
A look inside Google's Driverless car
The car uses
- Google Street View
- Software that combines input from video cameras inside the car,
- a LIDAR sensor on top of the vehicle,
- radar sensors on the front of the vehicle
- and a position sensor attached to one of the rear wheels that helps locate the car's position on the map.
The lack of intelligence is compensated for by providing more senses. Humans drive mostly using sight and can easily be distracted (or drunk)
Brilliant !
13
Thinking big ….
Internet Intelligence - distributed intelligence - an intelligent network
e.g.
You come home (the lights in your house turn on and off as you pass through) and flop in front of the TV. You wonder what to watch tonight….You whisper…"I'd like to see a cop thriller .. not too violent" …your internet enabled specs (or TV) offers you choice of suitable shows you have not yet seen….You've been feeling very tired of late, your smartspecs heart monitor sends data to a "Watson" diagnostic over the Internet and discovers you have a very rare, but easily treatable condition….
Intel has a Dream:
Iberdrola's Smart Grid is a step along the way (coming to Jávea soon !)
https://www.iberdrola.es/webibd/corporativa/iberdrola?IDPAG=ENWEBREDDISREDINT
The PC Plus article which inspired this talk can be found on:
http://m.techradar.com/news/computing/could-technology-make-your-brain-redundant-1040056
Chris Betterton-Jones
