Why the Turing test is not a good test of AI sentience/self awareness. For those of you that have a short attention span (ha ha)~ Reading the entire OT is not too necessary. Basically I make the case against machine self awareness, and why that the Turing test will not be a valid test for sentience and or self awareness (they are very similar but not the same). However if you are going to rebut or challenge my conclusions please read the entire thread to avoid time consuming and irritating 'redundancy'. Long version ; We have all heard the claims by the priests of secular science; "AI sentience nearly upon us". Those claims are believable to many researchers and scientists knowledgeable in the discipline of computer science. However I have serious doubts that arise from several areas including the metaphysical perspective*. For the period of the last 20 years the amount of computer RAM has increased exponentially by a level of 10 every 4 years! If computer scientists can sustain that (above) growth of RAM capacity* computers by the year 2029 will posses 8 million Gigabytes of Ram. That is roughly equivalent to the processing power of the human brain. My question is do you, the learned and kind members of PF feel that machine sentience will occur in the foreseeable future? Personally I feel the answer is no***. Even considering the constraints outlined by Moors law I feel that silicon based computers will increase in power enough to easily pass the Turing test in a few years. However sentience will remain beyond the reach of computers barring any unexpected break through for the foreseeable future. The Turing test is not a good indication of sentience for because if we could see the programming that went into the machine we could determine if there were code that would allow the machine to lie about it’s own sentience. If that code were removed, and the machine still claimed sentience well now, that would be something incredible! http://www.turing.org.uk/turing/scrapbook/test.html * By metaphysical I mean the question of a supernatural 'soul' or life force that is so far unknown to secular science. ** I highly doubt that the historical level of computer speed and capacity can continue at historical records for very much longer due to physical constraints, which includes material and the speed of light. If the light speed barrier and material constraints of silicon could be leapfrogged, the speed of computers and their abilities would be nearly infinite only limited again by the new materials and the quantum effects of say a quantum computer. *** Moore's law - Wikipedia, the free encyclopedia en.wikipedia.org/wiki/Moore's_lawPredictions of similar increases in computer power had existed years prior. ... At the end of the 1970s, Moore's law became known as the limit for the ... deep ultraviolet (DUV) light from excimer lasers with wavelengths of 248 and 193 nm — the .... Tech created a new speed record when they ran a silicon/germanium helium ...
I don't know about the turning test but about the last part; you need the code to stay and removing it doesn't really prove anything. All life on this planet has some base 'code' in it. You didn't have to learn to breath, make your heart work, taste, see, etc. You started with some base code and moved from there. A computer AI is going to have to work from the same point. You are right though; how do you tell if you're doing with an AI or something just programmed to say it's sentient? I think the key would be not to program it to believe it's sentient and see how it evolves past there. Give it some base (words, sentance structure, probably base emotion) and let it go past there. I needs to be able to learn and alter itself, being able to create new words, learn new concepts and emotional states, etc. Still how do you decide when it's a life form? At what point does it display sentience? If we let it evolve on its own it's entirely possible it could be sentient but not display it in a way we can recognize. Either way there's going to be a lot of argument when it happens!
I'm not really sure there's much of a point in quibbling over the differences between recognized sentience and theoretical pseudo-sentience. If computer AIs become indistinguishable from a biological intelligence, why shouldn't they be considered sentient? That's the point of the turing test. There's no good measure of sentience other than a comparison between a proposed sentient being and a recognized sentient being. Even biological intelligences are build from context and coding. Sentience doesn't arise in a vacuum, we're all affected by how the people around us have programmed us over our lives.
Yes, good thoughts Fling. I agree if we could view the program to verify that there is no code to tell the machine to mimic sentience and suddenly it says I am alive and I know I am me, I would probably tear up in awe. Then I may be fearful of a machine that knows its sentient and can theoretically think unbelievably, exponentially faster than our own remarkable brain is capable of! If it could replicate itself that may be very dangerous, talk about survival of the fittest VIA unbelievable evolution limited only by the speed of light and maybe not even that! Safe guards would be mandatory from the first inkling of sentience. That sounds despot like but it could mean our species survival! It would also raise very profound questions about God, which as a christian minister and theologian I would have to reevaluate a few of my dearly held christian axioms! Well IMO a life form would be different than a sentient life form. However if can build more of itself and or replicate etc we should recognize and protect it as such! I would say if it can do what the basic life forms we call life it should be said to be alive. We struggle to define biological life, is a virus life? Or is bigger bugs life, say bacteria? I would define even virus as life. Yes true we are now just accepting that some animals display rudimentary sentience. Some claim the entire earth or universe is alive. I never say never! Plants display electrical activity when exposed to some types of stimulus, such as if someone attempts to damage them. That has been known since the 60’s. So who knows? When in church2 i.e. in the outdoors, say when I am camping or hiking as I tire I look for the largest old oak or pine I can find. As I sit, resting against the tree, I can ’feel’ the life force that flows through it. it’s the same kind of warm almost feeling, when I look into friends eyes. Nice! That said and back on topic, I would say true sentience that we can understand would not be too difficult to verify, but difficult to prove to everyone’s satisfaction, as you so succinctly pointed out. Its time for a NEW Turing test or a Turing two. I think at the time it was created Mr Turing and his colleges may not have known how quickly computer power would increase. I remember asking a science teacher in 12t grade when computers would allow two way Dick Tracy communication as in his wristwatch TV/com thingie. He almost laughed at me and said it will not happen in my lifetime nor yours! It happened in his. Rev A
Please define the two, or clarify/expand a bit? Yes I have already stated that in the OT. The turing test can easily be defeated, we are almost there now. Have you read our posts? If I built a machine that was as you say indistinguishable from human how would you know that I had not simply inserted code that only mimicked self awareness? Or another way to look at it. When you lie down in bed at night that internal dialogue you have is your self awareness. You know you are you. However you could not prove to me that you were self aware. The difference between humans and a machine is that we can look at every bit of code that makes the machine work, talk and do tasks. We can not do that to a human. So If there is no code that tells a computer to say to humans it is self aware (that is highly simplified) then you would have a case for sentience. One needn't be religious or believe humans have a soul to determine the Turing test is flawed. I somewhat agree. There are no precise infallible tests to prove sentience. However, there are other tests of Sentience/self awareness. I somewhat disagree with you and believe that sentience, is instinctive in many animals. That does not include concionussess and or self aw not a learned process. (the words sentience and self awareness are similar but mean subtly* different processes etc) However there is some research that suggests that a learned process does have sentience (see below) and this link*; Self-awareness - Wikipedia, the free encyclopedia en.wikipedia.org/wiki/Self-awarenessSelf-awareness, though similar to sentience in concept, includes the experience ..... a b Mirror test shows magpies aren't so bird-brained life 19 August 2008. ... * I take the Wiki links with a grain of salt so to speak. I have been embarrassed before by using false or misleading information from Wiki.
You should open your minds eyes brother. Then you will see a wall in front of you clearly. Oh, there is a wall behind you too, its rather corrugated...hmmm. "Hey Rev!" you say to me, "I see four walls now, and a ceiling and a floor in this cube, no door or windows! ....How do I get out of this dammed cube?" the Colonel K fairly cries, the panic in noticeable his voice. The Rev and his metaphysical learned peers nod knowingly. "It's not a cube, brother Colonel K, its a box" I say sympathetically. "you are thinking inside that tiny box and have been since you were indoctrinated into the secular scientific paradigm of logical positive belief as a small child beginning in K-2 or K3!" Twilight Zone music gets louder, as the credits roll...... 'the end' With all due respect, if you (or anyone) claims 'secular science' does not exist. that is denial speaking to you inside that box, eh? Read a bit about the V. Circle and such things, free yourself from the constraints of secular science and the logical positive paradigm that has been manacled to you for decades. The ball and chain was not placed there your free will choice (If you live in the USA and a few other countries) but rather by the threat of jail time or worse. If you don't know what I am talking about I will provide the answer in my next reply. Hint; K.G. Rev A
already broke the back of the sentience with kurzweil emotions/empathy Moore's law is about costs silicon is already the dinosaur due to 'solitons, excitons, polaritons, etcerteron....etcerteron life. the 'soul' is the life of the mass nothing secular about it. self awareness is not a 'soul' but more like a coherance of..... morons dont comprehend what the 'life' of mass is, nor to concepts of what enables life to 'evolve' idiots believe it is magic, 'by the god(s)' them rules have already been proven incorrect with entanglement. but morons dont stay up on scientific comprehension to realize that. moore's law is about speed to production costs Moore's original statement that transistor counts had doubled every year can be found in his publication "Cramming more components onto integrated circuits", Electronics Magazine 19 April 1965: The complexity for minimum component costs has increased at a rate of roughly a factor of two per year... Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years. That means by 1975, the number of components per integrated circuit for minimum cost will be 65,000. I believe that such a large circuit can be built on a single wafer Moore's law describes a long-term trend in the history of computing hardware: the number of transistors that can be placed inexpensively on an integrated circuit doubles approximately every two years basic idiots dont comprehend much of what they read or write.
Genuine sentience is innate self-awareness. Theoretical pseudo-sentience is a non-self-aware intelligence that accurately mimics genuine sentience. I'm not sure the distinction is meaningful, but it could be made. If you write code that so accurately mimics self-awareness that it is indistinguishable from already-acknowledged sentience... in what way is that different from sentience? The turing test is not a technical challenge, it's a philosophical point made by way of a thought experiment. In other words, why would it matter the method of self-awareness, if the results are indistinguishable between methods? The fact that we can understand an artificial intelligence's reasoning--or, at least, the algorithms that represent reasoning--is rather irrelevant. Mystery and ignorance is not the root source of sentience. Irrelevant. Programming does not eliminate sentience. Human beings also think along predictable patterns, and likely according to a biological method not unlike high-level programming. We're just inefficiently programmed; by life experience, rather than by preconditioned commands. That does not preclude a computer from being self-aware, nor does it diminish the turing test as a valid thought experiment. Sentience develops with age, it's not an inherent property of biological intelligence. Babies, for example, are not self-aware until they develop more. That's akin to programming, because they learn a lot from the people around them--that shapes who they will become, and how they will approach problems.
Well, the turing test itself may be flawed. There was a counter argument to it called the "chinese room test". It ran something like this: One guy in a room, who communicates only through Chinese symbols with the outside world. So you in questioning this person via the turing test would give this person a tile with a Chinese word on it, and he would hand in the answer. From outside of the room, this is an intelligent conversation. From inside the room, you'd see that the person was actually only doing a type of symbol matching, not so much understanding the word you gave him, but taking that symbol, looking through a book full of symbols, and handing another back based on the rules in the book. So actually no conversation is taking place, just manipulating symbols by rules.
Computers are already sentient. They can power down when they overheat, etc. I think though the goal is sapience, which for some reason Sci Fi confused with sentience, which means to sense.
Turing test is hilarious really, so a computer is self aware when it can answer to a few questions? That reminds me when i was 6 years old and thought computers were alive because they said things to me ( like error messages ).
I haven't been on this forum for several years now and as you will notice my fore look of the world has changed. Regarding the future development of a Hard-AI, yes I believe we will achieve this feat primarily because I am confident that I will discover the underlying subtle mechanisms of intelligence. I am heading to graduate school for neurophysics and this will be the topic I devote the rest of my mortal life towards. First and foremost Hard-AI will not be programmed. Hard-AI will mimic the natural information processing capabilities of the human brain. The human brain's software is its hardware insofar as we have roughly 80 billion neurons arranged in a grid-like local topology with roughly 10 synapses per neuron that is constantly changing by stimulation of the physical world. The human brain is arguably the most complex object in the observable universe which raises philosophical questions of the feasibility of reverse engineering it. Simply accepting that consciousness cannot be replicated outside of a biological brain is a lapse of imagination. Surely if we are truly special creatures in regards of our intelligence we can replicate it and upgrade it. With regards to my current hypothesis of the emergence of intelligence I believe it to be a fundamental consequence of Ramsey Theory. Ramsey Theory is the study of complex systems that are sufficiently large enough that there are guaranteed substructures with a high degree of order. Contrary to the popular laymans beliefs order from chaos is not only compatible with entropy it is mathematically guaranteed. The laws of Entropy and Ramsey Theory are derivable from the same combinatorial axioms, they go hand in hand. Given the complexity of the brain consciousness cannot be avoided insofar as if every neuron is randomly firing then eventually structured information processing must occur.
