No Brainer. No Thinking Machine.
Remember The Emperor’s New Clothes.
The news tells me that Google has bought a small company for lots of millions, the company has never made any tangible solutions resulting in any profitability, so that’s a good track record in the software and web systems business!
It is October 2014 Newsnight BBC 2. The company is Deep Mind and they know that there are two roads towards A.I. Artificial Intelligence and on the journey to producing A. I. the processes may throw up new insights and methods. Right now there are two agreed starting points that experts see as the paths to generating such a leap in machine intelligence. Pre-programming is one way the other machine learning itself. Both are complex but the first involves mountains of work, the other may never actually produce intelligence but will produce functionality.
No one mentioned the Turing test in the television programme coverage. The Turing test is a machine’s ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human. I think this is a long way off maybe never achievable in its full meaning. We need to consider many aspects of the meanings of all of this.
Starting with Babbage: The Analytical Engine, The Difference Engine and I don’t believe its moved on in terms of intelligence; as repetition is the underlying, yet now, even more obscured key. Modern methods have dismissed the need for punch card instructions with software variables, but its the use of speed in the physical chips combined with repetition that provides the ‘known’ results, plus the capacity these days to keep sets of known results for additional use and then feeding these as the start of further operations. There is NO THINKING INVOLVED not by the machine.
So there are indeed two methods of approaching A.I. however in real terms they are confused as they are both linked. The first is to pre-programme (remember the punch card – punch card is now a programme) a known solution or sets of these are entered into a machine (or a further programme) to do calculations and execute results from. Basically its a program put into another program.
The second method is for the machine to ‘learn’ from these inputs and work out the results for itself.
Programming information into a known data containing system or network is where we are at currently; for example the Internet within which Google has algorithms that seek out itemised, strings or keyword identity tags to find associated results material. Web designers use SEO (Search Engine Optimisation) methods to match a Google search with a particular webpage. This may be biased to provide the largest sector on the subject, the locations or websites most used for it etc. Additionally results can be skewed by pushing advertiser’s pages as results for organisations that have purchased certain descriptive keywords (Pay per click). So ignoring Google’s inherent biases it’s the best example to date of a real world useful information retrieval system.
Google hired futurist Ray Kurzweil as Engineering Director in 2012. Kurzweil claimed that in 30 years, “humans will be able to upload their entire minds to computers and become digitally immortal” – he named this as ‘singularity’.
It is statements such as this that get attention and thus assist in book sales.
There is no foundation upon which to even begin to build the link from a human mind’s memory, let alone to obtain the functionality of the unconscious mind which makes up about 80% of what it is to be human; into a chipset.
The Film Transcendence has Johnny Depp uploaded into his computer and another the Spike Jonze film, Her in which Joaquin Phoenix falls in love with his A. I. operating system are money spinning derivatives of Kurzweilian thinking. These movies are Hollywood’s examples of fantastic assumptions. The problem is that such assumptions as films can become a sort of subliminal mythic belief.
Two films around similar subjects released about the same time and both actors have moustaches – a signal of intelligent design or just coincidental thinking. That may be a current enigma, however in the 1930’s there was a machine that started all of this off :
The German Enigma machine was a set of turning wheels to give cryptographic comparisons. Physical wheels turning, like the Babbage concept is not thought of as A. I. or some occult science because it has physical parts and real-world manifestations. What I believe is that because the functions of software is somewhat invisible it gains a magic quality. Yet it is still operating just like cogs and wheels; software inputs, and results are disassociated from thier physical connection to our real world and such mechanical operations, which in comparative terms is what is happening, is therefore hidden
We have got a confusion in our minds about the description of these operations and somehow believe (wrongly) for them to now have functions that are beyond our understandings. This is because the visible and intellectual concept of mechanical actions has been hidden within the operations of the inputs, outputs and invisibilities of software operations. However they are still ‘mechanical’ – mechanistic repetitive functions.
Roll on The Butlerian Jihad – a war against machines that ‘think’ and attempted to take over – similar to Terminator but as told in Frank Herbet’s ‘Dune’ which resulted in humans building better memories , better brain capacity and better predictive reasoning. Human computers evolving within human minds – conscious evolution, because it was necessary, well so it went in that book.
A. I. is a gathering, a collective of many things including; information correspondences, sensing, action – interaction, planning, reasoning, repetition and many more facets converging towards knowledge and then becoming predictive. But its not possible to deal here with all of that so a target upon a single theme will have more focus: words. Text and voice search methods: as words are the fundamental of Google search tactics.
So going deeper into this. Taking language, vocabulary, words as the descriptions of intelligent communication. Text, audio, word descriptions have meaning for us – language communicates meaning and understanding.
Programming information directly into a system for the objective of generating ‘apparent’ A.I. for text search results: means giving the data descriptors themselves extra – meta data – (data about the data). Tag descriptions of the meaning of each piece of data, i.e. like a dictionary and a thesaurus describing the words, their combinations and look up tables that describe these. All need adding to each and every word. Then taking these as the next input. Building this into a programme that then operates ‘on the fly’ almost instantly to ‘parse’ the word strings to glean the meaning of a Google search; which in return is then more able to produce a much more defined accurate result matching the needs of the user (internet searcher). A further step would then take these ‘meaningful programmes sts’ into other computational activities: the hope is that this will provide more advanced interactive systems with real meaning and outcomes. And not just more meaningless meandering Chat Bot type conversations.
Have you heard any of the mindless lyrics of Rap artists who are sort of rhyming word associations and stringing similar nouns together, often angrily.
Lets have a break before we have a mental breakdown: Tom Tom Club – Wordy Rappinghood .
Juxtaposing how we think with that of a machine puts all of this into a different perspective – do we posses concious thought from moment to moment? Can a machine? We just somehow know the meaning of words. Whether or not from such a data-bank, or the computer itself; such a ‘device’ would be aware of these meanings it associates to these resulting text, audio word descriptions or pictures is another thing altogether. But that’s exactly what the public might believe and the emerging A.I. industry have an agenda to keep things mysterious.
Indeed the root question posed above throws up insights into our own thinking mechanisms and our own intelligence. Do we really understand the meaning of our own brain function, our minds and conciousness? And if we don’t know that how then can we assess the supposed intelligent mind of a machine, a robot, or a piece of software?
It seems that much of our thinking is driven by automated function responses like anger, emotions that colour our responses, and many of our responses themselves can be unconscious responses which are learned over years from external inputs. Examples of this are put forward in the book ‘A Mind Of Its Own’ – by Cordelia Fine.
A baby as it grows leans from experiences, looking, hearing, touching, and language is thought to be subliminally learned before the child can speak. Then there is education. The abc is learned then words and their associated meanings and their representation as marks on paper, then we communicate these with others through speech, writing, typing. When we verbally communicate between ourselves. Quite often our discussions are coloured by automatic unthinking responses, most of the time we tend to be completely unaware of this! We ourselves can be mostly robotic!
Its difficult to observe this within oneself. So we can observe such things first in others. Continuous irritating repetitive statements. Bringing up the same subjects relentlessly. Picking up and repeating words and the statements of others. Words in vogue. Word viruses. Adding senseless adjectives at the end of sentences – you know!
Our responses in communication are coloured by many things; from certain knowledge and experiences, biases (often mingled with emotions, and our subconscious). We may have multiple minds (or selves) that come to the fore when we are at work, in the home, out socially with friends. In such environs are we constant or does our personalty drift due to the expectation of others? Are we constant and self aware from moment to moment? Our, our opinions our own, or put there by externals, the views of others, advertisers for example and then look at vested interests like a politician who does not believe what he says but is following a trend in the public’s attention that they want to be associated with, mostly to profit from. At other times we might argue an opposing view that we ourselves may have previously opposed. Also we have a stock of instant answers that are simply automatic responses. You can see its complex. A much more scruntinouse study of all of this is from: Idries Shah- The Commanding Self.
So I mention this to give a perspective upon machine intelligence as our own intelligent mind needs some understandings before we can claim such capacities in a machine. It is these surface operations of our minds, the learned automaton reactions that can most likely be programmed one way or another into a machine. But it is the other qualities that make us intelligent, human and emotional beings: I believe that such a combination will be extremely difficult to accomplish within an inanimate object, however it might exhibit superficial cleverness.
These methods already mentioned of programming for A.I. mean starting at the very basics with meta data meanings of everything: it’s one huge task and so due to the cost, efforts and resources required; nobody wants to go down this route. Shame, as I have always though this will be the most productive.
The second option is far easier in comparison and involves computing machines to observe, engage and gain knowledge over time of all moves and possible strategies, before putting all of this into a memory with a programme over-arched with this information. Good in games and gaming systems. Designed so as to act upon each event as a separate possibility and look up the result, thus to apply the correct option. Very much akin to the unthinking auto-mind that is the surface semi dreaming mind of many of us when we lose our self-awareness over to such automatic thinking: daydreaming. And don’t dismiss the need of this function. If we needed time to study the fact that our hand is resting on the oven hot plate instead of a pre-programmed instant reflex action to pull away, its unlikely that we would have survived so far. Instantaneous none thinking responses are required, but they are not the mindfulness that we would claim as our identity – our real self, but such responses are very much the type of thing that A.I. might well achieve. A computer is very good at this and once the strategies are all know it can do it at lightening speed, far faster than a human, but it’s a limited role focused upon single tasks. These may appear to be complex but in fact for a computer once set up and its ‘learned’ all the moves it makes are easy for it. This is a good starting strategy for A.I.
So Deep Mind is using a 1980s Space Invaders game to show what is meant by this. To begin with the machine plays slowly all options and loses all the time until it gains understanding through trial and error the best options. At first it is far worse than a human, later on it has no human competition, as all moves are known so it applies the best known to itself in its memory and wins every time and rapidly.
A human may learn all these moves in game playing but will forget them, the computer never forgets, but this is not A.I. Now, all of this posses questions; if humans had to learn everything by trial and error we would have died out thousands of years ago. Its most likely therefore that Darwinian evolution of subconscious responses had been learned and passed down the chain of living things from microbes to man.
Playing Space Invaders. The human may be distracted by hand movements, peripheral vision – eyes seeing other things in the surroundings. Also internal emotions that all influence the game playing and of course not being fast enough mentally nor remembering all strategies. There are other factors too: the human mind can be drawn off track by continuous internal thought, such thoughts even at the subliminal level will draw away focus; unlike the computer which has no such distractions but its probably those distractions that make us human. Such sidereal abstract thought in variable situations, takes up some of our surface awareness, we are multiple situation processing at some level – all of the time.
So a machine can be totally focused beyond human comparison and fantastic at Space Invaders or its potentials may used in other actions, like sorting out selections of text string results – which is what Google seek. But I maintain it will still be nothing like the A. I. as popularised into public understanding by the media and other influences. It will certainly be nothing like the flexibility of a human mind even with all its shortcomings of data memory retrieval, which the speed of a computer in doing such feats so impresses us into the belief of its intelligence; when its just moving very fast over its memory which unlike humans never fails it.
Yes its all exceedingly clever, useful and will become profitable. There will emerge many applications. Cars that drive themselves and other activities can be programmed in the same manner as option two above. It predictive observation, a baby learns in this manner, there are automatic responses and it leans from situations like the instant recoil from hot food, sharp objects and so on. External experience. Have you noticed how effective this is? It deeply programmes our automatic nervous system, much of which through Darwinian iterations is part of us from microbial single cell memory.
An electric turbine spins and generates electricity and through the interconnected wires turns the lights on. But nobody would claim that the turbine had intelligence. It is mechanically attuned to its task. I believe its functionality is in the same way as these emerging methods – like the Space Invader application – that are tagged as moving towards A.I. Such descriptions are faulty thinking.
The Space Invader and Chess playing machines such as Holmes and other applications are still, no mater how fast and efficient and beyond our capability, are not intelligent. A human cannot spin a turbine to produce electricity but we don’t stand in awe of its intelligence. So lets not be confused as these supposedly A.I. applications won’t be intelligent in our sense either.
The dark occult arts of supposed ‘intelligent’ computing minds of such ‘A.I.’ machines are mindless and robotic. Such qualities are useful for many future tasks and also extremely valuable, due to the human aspects that intelligent beings do possess, which they will not have.
Yes they will be very good for warfare; better drones etc without human emotions holding them back and no other such distractions, they will fulfil their tasks without ANY THINKING.
Deep Mind will initially be focussing upon using its insights into building better search results for Google. This has raised the issue:
When will machines emerge that think like us? A question posed already in the media showing the belief’ that this could be possible. When in fact it posses the question really of how do we think? and I believe that much of our thinking is unconscious and automatic and pre programmed and thus could be said to be machine like.
The TV show CAPRICA a spin off from Battle Star Galactica poses the way it could happen.
Many animals can do things faster than we can, and have capacities that we do not possess, yet they have not yet invented an industrialised complex society. Which may be the step required to reach towards where we are venturing now with these ideas of a post industrial, computer functional one.
A.I. its just a rapid habit!
So I believe that machine functionality is a programmed operation and in the case of such machines they do posses a very rapid capability: but its still in real definable terms a: HABIT! So My name for current computational machines in deference to Babbage is: A Habitual Engine. And a good name that shows to ourselves how habitual and automatic we ourselves are. The word habitual ought to be applied to the method of such ‘A.I.’ machines for a more corrective balance in understanding.
Continuing and further with exposition of the subject – would such machines have the capacity for separate motivation to take on unique new and challenging concepts to do things that interest it? What would be its motivation driver? what would be its reward? Ours it seems are all based upon attention. Attention to ourselves, our own self value, self worth, and you can see this everywhere; and bloated in celebrity, political and business people who have large ego drivers; its all reward seeking unto themselves. And we all do this to a greater or lesser extent.
Descartes: I Think Therefore I am, or is the question rather; I am because others think of me.
How would a machine gain such a quality of self and self interest or worse self obsession? Often in human terms this can be attributed to a fault, yet such a drive is linked to emotions of being and having an external identity that is admired by others. It is observable in its external forms such as: I have followers, I have Facebook friends, I have others that like me: thus I am real. Is that phone ringing for me, do I have any new messages, is anyone looking at my messages, my texts, my TWEETS?
A machine has no such requirement for responsive pats on the back, even if it were possible to imbibe it with such a protocol, which I doubt. So I don’t think that we will lose control over such machines as they will not run amok unless programmed to do so. I believe this is so because its seems that our ‘intelligence’ is driven by rewards that give us identity and attention. This is an intangible and not a functional thing that could be put into a machine to then enable it to be intelligent like us, because then it will be as dumb as us!
Solving cryptographic crossword puzzles is also habitual! We are rewarded for doing so by thinking how bright we are, when we are really using memory and associations, just like an intelligent computer.
The boss of the deep mind company said on the BBC TV programme Newsnight that the resulting technology and derived devices would be sensibly used and not for military purposes! I think the very opposite. Its that driving force and its funds that will indeed drive such technology forwards. And they the military don’t want thinking machines, it is the very fact pointed out herein that these machine will NOT THINK that is the factor of value to certain people. They don’t really want thinking emotional and conscience driven soldiers.
It can’t be stopped, you cannot un-invent an atom bomb. So its coming. Its predictable that we will see none emotionally driven, single focused (dare I say it robotic) singularly fixed ideologically (a program directive) driven combatants. A rapid robotic unquestioning mechanical soldier is just the job. Somewhat like its human comparison: the idealogical and mindless IslamicState robotic thinking lunatics.
To mollify us the boss of Deep Mind assures us that Google is an ethical company and thus his technology now bought out by Google is in safe hands. Well I think that the military complex somewhere will in anycase be simultaneously developing this stuff. And lets not forget the fact that the Snowden leaks have already informed us that Google tips off anything that the secret NSA people want – already.
A.I. Attack Droids seem to me to be a real outcome. The future soldier, metal boots on the ground. Robotic Terminators! But I doubt they will have the self attention seeking characteristic of we humans, and so will obligingly stay within their programmed functionality.
Deep Mind that’s a curios name tag and one that gives an impression of a deep intelligence when in fact its clever, very clever, but its still just mechanical actions at very high speed. We should know it for what it is; a mechanical function and mechanical things are mechanical and don’t really have any integral deep thoughts and no mind. We need to be careful we don’t miss comprehend all this. These devices imbued with ‘A.I.’ will be amazing, and useful but they won’t be self aware and thinking, they wont object to being switched off. Lets not classify it as thinking, intelligence, and not really the A.I. we are being conditioned to expect. We are being manipulated by popularised simplistic thinking (another thing we can easily fall prey too) and thus wrongly believing these emerging ‘A.I.’ applications puts extra value to all of this and we then ad ‘wonderment’ to it because we cannot do it but don’t be confused: what our eyes see and our ears hear does not mean such things have internal intelligence nor sentience.
Humans have a lot of real-time simultaneous mental functions, and often the faculty of emotions, guilt anger, humour and – that thing called conscience.
A conscience comes from the word consciousness it means to be aware. We have simultaneous running mentalised functions: separate awareness of our actions, even of our own thoughts.
An advanced meditation practitioner for example can observe their own thoughts. And become aware of an identity within themselves of a faculty known as the ‘observer’ which is able to see and passively observe internal thoughts as they rise, interconnect and fade. This faculty all humans have and it shows that we have a separate consciousness to all of the mechanical automated response mechanisms that have been programmed into us from childhood by environment, education, upbringing, beliefs, emotions and all that. This observable separate identity apart and distinguishable from the thoughts we have – this awareness – is what separates us from the animals.
No computing machine even with amazing abilities and speed can have such an internal capacity as that described above, if it could be given to it it would probable be less functional as all those attributes would slow it down just like us. It would become responsible for its actions; if fully aware, (as often we too are not!)
What A.I. does show us is how we delude ourselves with descriptions of a thing like A.I. and indeed how for most of our day we are distracted unthinking automatic and robotic in our attitude responses and our thinking. For the most part of our day we can be involved and mentally lost in an internal world of endless thoughts. A dialogue with a continuous talking to ourselves, incessant inner chatter. This shuts out that awareness of our inner deep self that is always observing and sometimes butting in when its essential: like the application of our conscience, having free will. The ability to distinguish wrong from right.
Conciousness is a word derived from a state of being concious, being aware. Now all of this emerging story about artificial intelligence professes to be also an acquisition of conciousness that sort of just comes along with A.I. but what I have attempted to point out here is that real conciousness is a very different thing. It is also evolving. And we should use these insights from observation of external advances in clever systems – Habitual Machines – to make us more aware of our own state of Conciousness.
More Mind Stuff:
Our Dreaming Minds (SGO)
Comments, interest etc
THIS POST RAISED A FEW INTERESTING FOLLOW ONS:
and a lot on the Evidence for an Electromagnetic
Field Theory of Consciousness.
From Dr Joanna Bryson
University of Bath
Dr Bryson was on the BBC Newsnightprogram
and was not given time to offer her thoughts on
the questions and ethics of A.I
She suggests looking at her page:
And continuing monitoring the subject.
FROM J. HADLOW
Interesting. Machines are machines – software or not, as you say.
Maybe this is why machines are not conscious: http://www.surrey.ac.uk/qe/
Evidence for an Electromagnetic Field Theory of Consciousness -Johnjoe McFadden
I read of a theory a long time ago that suggested that electricity resides in the fields aroundconductors, electrical components etc, and what we see flowing in the wires is the reaction to those fields existing on the outside of the conductors. In other words, it is the reverse of current beliefs that electricity travels in the wires. It doesn’t in this theory. It travels outside the wires, but we detect something in the wire, because our machines for detecting it, create the conditions for measuring it there. I.e. because we “interfere” with the wire when we test for electricity, the “interference” causes our measuring instruments to register something. A bit like waving a coil of wire in a magnetic field. It must produce electricity.
I wish I could remember who the author was of that, but it fits with the above link, and if we take that a step further, then if that field is conscious, that explains why software or other machines are not conscious. I.e. they don’t have that external field of consciousness associated with them.
REPLY: It was Eric Dollard who describes electricity that in fact travels external around the wires, not in them and thus creates EM fields flowing both ways and even accumulating power: THE FOUR PHASES OF ELECTRICITY.
Another from J. Hadlow:
I just found that article I had buried for years on electric field theory.
The man (Ivor Catt) might still be alive. His website also has one of those sob stories of rejection by the establishment (I think you told me a few stories about that too!). He was involved in the early days of computing, and did some important work for some big companies of the period.
Dollard probably got it from this fellow Catt who has been banging on about this since the 60’s.
See the links, Part 1 & 2 here: http://www.ivorcatt.co.uk/
I haven’t had time to digest this, as it only back came to mind yesterday, but it seems relevant to the field theory of consciousness amongst other things.