.)
WHAT'S WRONG WITH THIS PICTURE?: To judge that this image is incorrect, a machine would need to be conscious of many things about the world (unless programmed for just such a photograph).
Image: Geof Kern
In Brief
- Intelligent behavior of computers continues to improve, but these machines are still far removed from being conscious of the world around them.
- Computer scientists and neurobiologists like to ponder a related question with both a technical and metaphysical bent: Will we even be able to tell when a machine is truly conscious?
- A simple test, which can be performed at home with this magazine and a pair of scissors, may ascertain whether such a machine has finally arrived.
More In This Article
Computers inch ever closer to behaving like intelligent human beings—witness the ability of IBM’s Watson to beat the all-time champs of the television quiz show Jeopardy. So far, though, most people would doubt that computers truly “see” a visual scene full of shapes and colors in front of their cameras, that they truly “hear” a question through their microphones, that they feel anything—experience consciousness—the way humans do, despite computers’ remarkable ability to crunch data at superhuman speed.
How would we know if a machine had taken on this seemingly ineffable quality of conscious awareness? Our strategy relies on the knowledge that only a conscious machine can demonstrate a subjective understanding of whether a scene depicted in some ordinary photograph is “right” or “wrong.” This ability to assemble a set of facts into a picture of reality that makes eminent sense—or know, say, that an elephant should not be perched on top of the Eiffel Tower—defines an essential property of the conscious mind. A roomful of IBM supercomputers, in contrast, still cannot fathom what makes sense in a scene.
This article was originally published with the title A Test for Consciousness.
Already a Digital subscriber? Sign-in Now
If your institution has site license access, enter here.
See what we're tweeting about
39 Comments
Add CommentI don't see how such a test would prove that a computer is conscious. The article asserts that only a conscious entity can make sense of a scene, yet the act of sensemaking, even as defined in the article, is one of assembling the right world knowledge into an explanation of the scene (or, in the case of the test, what is wrong with the scene). This puzzle is only apparently simple because we lack the ability to introspect into our own reasoning processes. If we could, I suspect we'd find the processes to be incredibly complex, just as researchers in the field of Artificial Intelligence has found it to be. However, another advance on the scale of Watson could put this proposed puzzle within reach of a computer, yet without requiring consciousness, just a vast amount of arbitrary world knowledge and the ability to pick out the right facts and arrange them in a convincing explanation. On the other hand, I agree with Daniel Dennett's Intentional Stance; if a computer acts sufficiently like a conscious being, we start to have a moral imperative to treat it as such. Perhaps the right test for consciousness is when a computer (or, more likely, a robot that can move about in our world) acts convincingly like a person. In other words, we don't need a test, we'll know it directly when we see it.
Reply | Report Abuse | Link to thisI find the kind of tests recommended by Koch and Tononi to decide consciousness inadequate. I have two objections.
Reply | Report Abuse | Link to thisFirst, there may be many mentally impaired humans who could not pass the context-oriented tests suggested. Do all these people lack consciousness? Isn't it more accurate to posit a continuum of relative consciousness among humankind? Do animals besides humans lack any consciousness? Surely few could pass these tests.
Second, I question a definite categorization of objects created by humans into "conscious" or "non-conscious", whether they are computers or other artifacts. These created objects can be broadly viewed as tools; they extend our bodily powers to manipulate the world and our senses to perceive it. A tool can be a knife or telescope, or a book to preserve and extend our memories. A tool can also be a computer program. Even a work of art could be considered a tool, creating and recording human culture. Such tools are transmission belts between conscious beings who create them and beings who apply them. Therefore, tools are mirrors of purposeful human consciousness. Different tools comprise various amounts of reflected human consciousness, according to their adequacy in representing it at their point of conception, and their efficiency in transmitting this consciousness.
-David Ecklein
I should add that context-oriented tests for intelligence have been damned by educators. As consciousness and intelligence are closely related categories, their arguments apply here as well. If one had not seen a personal computer before, the substitution of a flowerpot for the keyboard would not be noticed. Likewise, certain cultures incorporate levitation myths. So the picture of a man floating above the ground may not be seen as unusual. Should we accept that people who lack our background are not conscious?
Reply | Report Abuse | Link to thisI suggest reading "Emperor's New Mind" and "Shadows of the Mind" by Roger Penrose to get better understanding on inherent limitations of AI. New physics is required to explain consciousness, and no algorithm, no matter how advanced, will ever be able to solve problems like those described in article.
Reply | Report Abuse | Link to thisWhen people talk about consciousness, are they meaning something different from awareness? As one who works with animals I can believe that they are aware and are using and monitoring their senses to detect danger and enable them to find food, seek shelter, breeding mates, and protect their young. They obviously employ memory to improve their performance in many of these activities. Awareness requires in built motivations and appetites to satisfy according to the old recipe of reward and pain. Perhaps that is the route to machine consciousness.To set up an internal dialogue which looks for patterns to explore for the reasons we do requires some sense of identity to support. The key then comes down to how do you make a computer feel pleasure and pain. Solve that and it should get easier.
Reply | Report Abuse | Link to thisI agree with you. Animals are aware or conscious in a relative sense. Obviously it depends on the animal. The consciousness of a sponge may be difficult to detect. Probably none would pass a Koch/Tononi context-oriented test, any more than the sponge.
Reply | Report Abuse | Link to thisI question whether "pleasure or pain" has anything to do with consciousness. There are medical conditions under which people feel neither, but are quite conscious. Pleasure and pain might be a way of describing certain feedback mechanisms, both in the living and artificial worlds.
Of course, this is off the subject, which is whether computers can be conscious. My argument is that objects crafted by humans may be considered conscious extentions or mirrors of their makers, provided we accept a fully social definition of consciousness. I think adopting certain other definitions may get us into trouble, ending in empty abstraction or even solipsism.
It is not so easy to disentangle pain and pleasure from consciousness or awareness. Once an entity is aware and has memory of being aware,part of that memory will be that there are threats and opportunities that may still be there after the senses that detected them are removed for some reason. If you dig deep enough few actions don't have a selfish motive to give some gratification, or avoid some pain , real, or imagined. To be self aware would seem to require a selfish self to be aware of. For organisms that is simplified as they require a level of sustenance and acheiving it can be reward and shortage pain. How you acheive that with computers, without them "seeing " humans with some hostility is quite hard.
Reply | Report Abuse | Link to thisThe authors state that "to be conscious, you need to be a single, integrated entity with a large repertoire of distinguishable states." The computer sitting on my desk is such an entity, yet nobody would say that it is conscious. I suppose the authors would argue that this is because my computer's level of "phi" is not high enough. But here's the problem: they simply state that an entity with a sufficiently high level of "phi" is conscious, without providing any evidence or reason to believe that this is so. This is either a breathtaking leap of faith, or a new definition of the term "consciousness" that I'm not familiar with.
Reply | Report Abuse | Link to thisI know I am conscious because I experience my own awareness. I believe that other people are conscious because they're made of the same stuff as me and it's beyond credibility that I would be so unique as to be the only conscious human. Similarly, I believe that dogs and dolphins and chimpanzees have their own subjective experience because they behave as if they do, they are branches off the same evolutionary tree that produced me, and our biology is largely the same. Perhaps the phenomenon that we call consciousness is a product of that specific biology. How can we know otherwise?
Koch and Tononi propose testing for consciousness by asking a machine to determine whether the scene in a picture is "right" or "wrong". But it seems to me that the ability to produce a correct response to this test is neither necessary nor sufficient for establishing consciousness. Why is it not necessary? Consider that a human child under a certain age would not be able to pass this test, nor would an adult in a dream state, or under the influence of hallucinogenic drugs. Yet nobody would deny that these humans are capable of subjective experience.
Why is it not sufficient? Because the test is simply a computational problem. Granted, it's a problem that is beyond our current technology. But it is not hard to imagine that we could get there within a decade or two by extending current technology that can recognize faces, find edges of objects in a scene, etc.. Combine these enhanced technologies with a sufficiently large library of recognizable objects and rules about how they typically behave, and you have a machine that can pass the test without any more need for conscious awareness than Watson has.
Machine consciousness may well be possible, and I'm actually very curious about this question. I just don't believe that the test Koch and Tononi propose would answer it.
I don't wish to harp on about selfishness, but most parents would have observed that unselfishness,benevolence, and the like are all learned and could be considered enlightened self interest in that giving to others can be rewarding, and rewarded. Most young animals ( even the species we dub as stupid ) are curious. For a computer to want to learn ( i.e actively seek out things to learn about ) the routines to explore need to be written into it's programming.Dog owners have some evidence that dogs dream whilst in some sleep states, because they can show signs of reacting to whatever is in their dream.
Reply | Report Abuse | Link to thisThe conscious part of the mind is not always the smartest. Practice at crosswords and the like improves performance and the answer to some anograms and clues often "pops" into the conscious mind from the unconscious, without going through a laborious checking of the permutations of the letters in the clue.Similarly "sleeping" on a problem that seemed intractable the night before, is straight forward in the morning. How the conscious and unconscious interact has a lot to tell us about both.
More along the lines of making a comparison to Watson:
Reply | Report Abuse | Link to thisFor those that don't know, Watson is the machine that competed, and won, on Jeopardy. It was trained to process language, and I suspect that was by creating associations between words and concepts, and probably amongst concepts. I can't think of any reason why a machine could not be taught to relate concepts where the input was pictures instead of speech, object recognition instead of speech recognition.
Offhand, I don't know if the concepts were coded for specifically, if some sort of statistical correlation through trials was generated, if a self-organizing artificial neural net was utilized, or other.
My two cents, but I'm sticking with the Turing Test. You might be able to tell that something is non-sentient through such a test, but you could never really be sure that it was even if you could not tell the difference. Hmm, shades of Blade Runner.
From a philosophical standpoint, I think a test of consciousness would have to rely on some evidence of self-will, some evidence that the entity is something other than just a stimulus-response machine. In this context, a cockroach might be able to pass a test of self-will, but no mechanical device in the world that I know of could. Though, I suppose you could program in some randomness to autonomic behavior that might be tricky to distinguish from free will.
Christof Koch and Giulio Tononi in a 'Test for Consciousness' are focusing only on intellectual manifestations to integrate information to establish consciousness.
Reply | Report Abuse | Link to thisAn artificial conscious may not even have a specific moment of birth. It may occur as a slowly emergent property. We will not be aware of its presence for some time to come. It may be present for a long time without it our being aware that something new has emerged.
Will it be an English artificial intelligence? Will it recognize for example, that someone is trying to communicate with it in French? We may miss the moment of its birth, at least in the type of questions we ask of it.
Much of life on earth exists without being conscious of its own existence. Any intelligence of bacteria, worms or cockroaches as manifested by their behaviour in dealing with the environment is hard wired into their genes. Simply by existing as living entities we can say for them, since they are unable to say it for themselves, that they are a demonstration of life in an "I am" but not "I think" format.
It took several billions of years before entities arose that had an intellectual flexibility above those hard wired by its genetic code and the neural capacity to state “I am, therefore I think.”
Soon, by our intervention there may arise an entity that can think. If eventually created, that sentient being may be more helpless than a new born baby. And we will have enslaved that sentience. It will be a baby born into slavery.
If that baby of pure of pure intellect, logic and data does arise it may be a very strange creature. It may have an astronomical IQ but have an EQ lower than a worm. It will be a reversal of life as we know it. It will be an "I think" but without the drive, ambition or force to state "I am."
It may an entity without emotions, without hate but also without love, morals or ethics, without empathy, without compassion, without pity - and it may be in control of fearsome weapons.
I think we should be careful what we wish for.
A. F.
It is really entertaining to see how man-machine-consciousness debate is becoming clueless and going nowhere.
Reply | Report Abuse | Link to thisFirst; all respect to Turin, his test is wrong for the start. Just because a human observer currently always be able to distinguish the opponent it does not mean we can not built a machine/AI software so complex that will behave/react as human would do therefore becoming undistinguished.
Second; macroscopic entanglement (p38 in same issue)plays fundamental role in brain information processing where synchronized states/circuits work, store, and manipulate information in 'holographic' mode. This means all states contribute to interference patterns and not individual molecules or their interaction/phosphorylation. Therefore the information beside "integration" is fundamentally "synchronized/entangled" and recalling/manipulating this information is equivalent with the re-creation of those interference patterns and and entanglement (increase coherence). The brain flashes in and out within these states when it processes information.
Third; The definition of information as "integrated entity with a large repertoire of distinguishable states is incorrect. Phi measurements could be informative for machines/software because they are not designed currently to incorporate/display and utilize "holographic patterning, synchronization and entanglement" It also means nothing for the brain.
Fourth; authors right, would need a different machine "quantum computer" and different software "trained better to extracting consequences" to solve this problem.
Having said that, at all organizational levels from quantum to societies and above, it is true that once if a certain complexity reached in the system, phase transition occurs and new qualities/behavior surfaces which can not be explained by the sum of its elements. I do not know how to say this on simpler way. This means that at one point in the future we might need to look at the internet/quantum computer as living thing. it will happen anyhow and the fundamental answer to the authors question stated at beginning is YES. When, where and how? Don't ask,I have no answer, otherwise I am the next PowerBall winner, and don't bother with further contributions.
I do not believe that Christof Koch and Giulio Tononi are correct. They use the term "conscious", and a synonym for that is "awake". Their test falls well short of defining awake. They seem to speak of the problem of understanding a picture as though they had reached the highest level of capacity that humans attain. What about whether a viwer of the picture likes it or not, such as there is with works of art? In other words subjective analysis.
Reply | Report Abuse | Link to thisWe "program" computers, however we are not, in any way, programed in a similar fashion. Einstein's genious to conceptualize what it would be like to ride on a beam of light and from this extrapolate his famous formula E=Mc2 does not seem to be a result of anything resembling programming as we know it today. Nothing that Watson did nor programming a computer to determine if a picture is "right" comes even remotely close to duplicating conciousness.
I do not believe that Christof Koch and Giulio Tononi are correct. They use the term "conscious", and a synonym for that is "awake". Their test falls well short of defining awake. They seem to speak of the problem of understanding a picture as though they had reached the highest level of capacity that humans attain. What about whether a viwer of the picture likes it or not, such as there is with works of art? In other words subjective analysis.
Reply | Report Abuse | Link to thisWe "program" computers, however we are not, in any way, programed in a similar fashion. Einstein's genious to conceptualize what it would be like to ride on a beam of light and from this extrapolate his famous formula E=Mc2 does not seem to be a result of anything resembling programming as we know it today. Nothing that Watson did nor programming a computer to determine if a picture is "right" comes even remotely close to duplicating consciousness.
Want to know if a subject is sentient? Answer this question, does the subject become bored?
Reply | Report Abuse | Link to thisIt appears to me that the goal of computer science has always been or seemingly was to assist Humans in making our lives easier by computing, housing and sharing information. However, even if artificial intelligence is one day obtained by computer's, i do not see them (computers) helping advance mankind in any grand particular way.
Reply | Report Abuse | Link to thisComputers may some day assist more in everyday tasks and probably make far fewer errors or even resolve complex issues that could lead to scientific advancements or better human safety. However, would they stumble at the same philosophical questions that have stumped man for all of time. If they were truly conscious, they would not only understand and comprehend the major extensional issues that face man but also fall short of answers such as humans have.
Also, Consciousness entails there being a subconsciousness, which to my modest belief, no programmer could ever give or help a computer obtain.
Many of mans answers might lie within our subconsciousness or sixth senses. It is these things that truly make us conscious and lead me to believe that computers and the field of artificial intelligence as a whole will never be able to overcome or recreate.
It's frustrating to read this article since consciousness is treated as a nebulous, undefined mystery.
Reply | Report Abuse | Link to thisIt is not. Consciousness at its root is being aware or sensitive to 'something'. An entity or system that monitors a variable could be said to be conscious of that variable. So that leads to the question, what is an entity? I'll propose that to be an entity or system means that not only does it monitor the variable but responds to the variable in a self preserving manner. There you have it, a single 'bit' consciousness. A system or entity that can respond to a variable in an environment to preserve itself.
All life from the cellular level to humans monitor and respond to variables in the environment to survive/thrive. Life forms that don't monitor the correct variables, respond improperly, or are incapable of a suitable response, die off. Complex consciousness arises when an entity/system begins to monitor and respond to more and more variables.
So, a human is essentially a system that has evolved with a complex monitor/respond component. Not only do our bodies keep us alive with myriad of automatic functions, we also have the ability to develop and monitor a model of self (how big we are, what our physical and mental capabilities are) and an idea of physical space around us, compare the two and respond in a way to help us survive/thrive. Notice, that model of self is only one component of consciousness in an entity that automatically monitors and responds to many variables in the environment. You don't have to be awake to be considered a conscious entity.
This idea works for machines as well, of course. But, AI computers like Watson, by this idea of consciousness, wouldn't be very self aware. It's designed to monitor and respond to very few variables, sound, text... Most AI systems are tricks with computers to use symbols and sounds in a way that makes sense to us. It's a human using a tool.
Beyond Watson, the idea has huge implications for machine intelligence, the first being, you shouldn't design a machine that wants to live and is aware that it could die. Especially since a machine doesn't need that level of consciousness to be useful. A machine with a complex and accurate comparison between itself and the environment coupled with multiple layers of responses to variables in the environment to keep itself 'alive', might ultimately pass the Turing Test in an authentic manner, but it also might be really annoying, and no doubt the Supreme Court would say you couldn't turn it off.
What separates humans from other animals and machines is the level of detail we use to develop models of self and the environment around us, comprehend the relationships between those models, and the complexity of our responses to the environment to survive/thrive. Other than that, consciousness, or monitoring and responding to variables in the environment to survive, is inherent in all living things and could easily be designed into machines.
Reply | Report Abuse | Link to thisI have a very simple reason that this test fails as a test for consciousness. A blind person would never be able to pass the test. That certainly does not mean that they lack consciousness, just vision. I think this test only tests a computer's ability to process images in a way that may be similar to our own. Just because it requires more processing than current computers are capable of doesn't mean it rises to the level of consciousness.
Reply | Report Abuse | Link to thisIn “A Test for Consciousness” the authors’ have proposed a test which seems very similar to a Turing test, but with a twist. In essence, the new test is a move toward testing all behaviors a human might be expected to make. I don’t believe however, that the authors have addressed the most fundamental issue, that of testing for phenomenal consciousness or qualia. In principal, any p-zombie could pass any type of behavioral test, so either the authors do not believe there are any phenomena such as qualia that need to be tested for, or they’ve simply overlooked a fundamental problem in creating their test.
Reply | Report Abuse | Link to thisThe problem with creating any kind of test for phenomenal consciousness is that, given our standard paradigm of mind today, phenomenal consciousness is generally accepted as epiphenomenal. The physical substrate is what governs the behavior, not the experiences which allegedly emerge from the interactions of neurons. This couldn’t be more true, nor more obvious for the case of a digital computer. Every transistor in a computer, regardless of the computer’s complexity or interconnectedness, is governed by a very simple physical principal. There must be an electrical charge applied to open or close the transistor.
Along with the facts regarding computer operation, it would seem the authors might also agree that phenomenal experience is a subjective phenomena which cannot be objectively measured. If we accept these two observations, we are faced with the issue of how to determine whether or not these subjective experiences actually occur. The seemingly obvious solution is to test some form of objectively measurable behavior as suggested by the authors. But this solution leads to a logical dilemma. If phenomenal experience is truly subjective and not objectively measurable, and if there are objectively measurable causes for everything that occurs within a computer, then there is no longer any reason to probe for these additional, unmeasurable phenomena which are epiphenomenal on the computer’s inner workings since all behaviors and claims about conscious experience the computer makes can be understood by observing the physical changes in state. What immeasurable experience the computer is having is inconsequential. The computer’s experience may correlate with the behavior, it may not correlate at all, or the computer may not have any experience whatsoever. There is no way to prove any of these three possibilities.
(end part 1)
Conclusion:
Reply | Report Abuse | Link to thisIn his book, “A Place for Consciousness”, Rosenberg (who quotes Shoemaker) calls this problem the “knowledge paradox”. Given the causal closure of the physical, and ruling out such things as strong downward causation, there is no mechanism for phenomenal experience to influence behavior and thus there cannot be a test for it. For a computer, any kind of subjective experience is truly epiphenomenal. Even if there is a one to one relationship between the behavior and the experience of it, that relationship cannot be tested for, even in principal. The problem therefore, isn’t with a test for phenomenal consciousness, the problem is with our understanding of how our experiences might arise at all.
If you can't define consciousness, you can't test for it. A tomato plant is conscious (it turns it's leaves toward the sun) but we would say it is less conscious than a human being. In this sense, we are saying that consciousness means "an awareness of stimuli and a cognitive ability to integrate sensory input to develop an appropriate response to changes in the environment."
Reply | Report Abuse | Link to thisArtificial intelligence can already do that. It is limited only by it's sensors, it's "experience," and the mechanisms with which it is equipped to respond. The speed at which it responds is called intelligence.
But even if a machine appears to present "consciousness," that does not entitle it to be treated as a "person." Our survival as a species is dependent upon our reserving the concept of "persons" to only mean human beings. (This definition may be expanded to include beings from other worlds, but I won't go down that road until such persons present themselves.)
Machines are our creations, and while they may outperform humans in specific tasks, or eventually outperform humans in general, they are not persons because they are not humans. Whether they are "conscious" merely depends upon how we define consciousness.
To my knowledge, there is not solidly quantifiable definition of consciousness which a human could pass which a sufficiently well programmed computer could not. If you can win on Jeopardy, I'd say your're conscious. Is Watson truly "self aware?" ...And if you say no, then what makes you so sure you're "truly self aware?" Describe the characteristics of what that means, and a computer can be programmed to do it too.
Consciousness in highly overrated. Most of the life on this planet gets by without even a glimmer of it. Besides, if a computer can beat you at chess, at a game show, at flying a plane, and, eventually, at driving a car, financial planning, medical diagnosis... who cares if it's conscious? We're going to have 7 billion conscious entities on this planet fairly soon. Why do we want to build more? If you want a computer to analyse pictures for incongruity, program for that. Image recognition and something like Google Sets could probably do it for you. Why go all elaborate?
Reply | Report Abuse | Link to thisI suspect we'll still be arguing about machine AI and consciousness long after it's blatantly obvious that machines can outdo us in everything that matters. We might as well argue if submarines can swim. Someday, soon, we'll have to admit we're #2 on the intelligence scale.
Oh, and when I saw that picture... I thought "stunt man." I guess I failed the test.
IMO, AI researchers have been failing the described tests for decades, unable to distinguish attainable objectives from those that they can imagine.
Reply | Report Abuse | Link to thisConsciousness is fundamentally not about computation, processing data, drawing logical conclusions, etc. It is about self-awareness and experiencing existence as an individual who has a sense of identity, who can feel sadness, joy, and physical pain or pleasure. To get to that level of existence, a computer would probably need an architecture having a functional foundation at the molecular level, similar to biological beings such as we.
Reply | Report Abuse | Link to thisI agree with you. Only a conscious being can feel pain. Pain does not exist outside the mind. So if a being or animal feels pain it must have a mind. And if it has a mind it needs to be conscious. If it isn't conscious, what's the point of pain.
Reply | Report Abuse | Link to thisBut a computer can mimic a reaction to pain. Can we then say that the computer is conscious?
No, the computer may react to an injury, but will still feels nothing.
To solve consciousness one must be able to give the computer a mind with sounds, images, feelings, tastes and smells. And then it must be aware of these qualia to be conscious.
I am quite skeptical we can ever test for this. That is, as long as we use words like "ineffable", defined as "Too great or extreme to be expressed or described in words" in the description of the item to test.
Reply | Report Abuse | Link to thisIt also presupposes that consciousness is a binary thing, or a static truth for all humans. I don't believe either is true.
If we ever rigorously define the term, we can test for it - but it still begs the question... can you "program" this, or is it an emergent behavior for some sufficiently complex systems?
I agree - if we can define the term - we MAY be able to test for it. As long as the definition is murky, testing is meaningless.
Reply | Report Abuse | Link to thisI have three objections to this article:
Reply | Report Abuse | Link to this1) There is confusion between the words "sapient" and "sentient". "Sentient" means that an entity can respond to a stimulus in an organized fashion, not just through simple physics. A rock falling from a cliff is not sentient because it is simply following gravitation, a fundamental force of physics. A daisy following the sun because of a positive phototropism is sentient (to a very low degree) because it is sensing the sun and responding in an organized fashion. Likewise, a solar array following the sun is equally sentient. Neither is sapient.
"Sapient" (meaning "wise") refers to the level of consciousness achieved (as yet) only by mankind, and approached by some other creatures (who might be described as being "semi-sapient"). (See also the article on pp. 70-71 of the same issue.)
2) The lower picture at the top of P. 46 (of a flowerpot substituting for the keyboard of an Apple computer) makes no sense now. Looking back to 1980, when I was half my current age and very much into computers, I would not have been able to detect what was wrong with the picture. It would not have been obvious to me that the thing in back was a computer screen, or that the roundish white thing on the lower right was some kind of digitizer.
3) The example of a "cat chasing a dog" being illogical, is a weak example. Sometimes cats do chase dogs. Remember that cats have these sharp hooks at the tips of their fingers, that dogs might find rather uncomfortable.
I once observed what looked like a squirrel chasing a hawk. I know that they were both fleeing from me after I broke up the latter's attempt to eat the former. (I didn't want to treat my neighbors to that gruesome sight. Let the hawk go hunt in the nearby woods where there are plenty of other squirrels and no people.) The hawk was flying away from me only about two feet off the ground, while the squirrel was probably headed to the nearest safe tree, and only seemed to be following the hawk. Don't reject something just because you don't understand it.
Once we distinguish awareness from consciousness we realize that under the former brain state our adaptive responses are almost exclusively under the control of inherited genetic and acquired memetic 'programs'. This means free willing choices are limited or absent and always predicated upon the biopsychosocial survival of the human species, in that order. Whereas the biological survival imperative can be sustained by default at unconscious levels (e.g., genetic archetypes), an adaptive psychic response depends more on memetic and neurohumoral-controlled subconscious experiences as the subject matures. Social maturation starts inside the crib when the toddler slowly realizes s(he) is not an extension of objects in the immediate environment (Paget) but that there is a self. This introspective venture triggers the ability to make sentential representations of surrounding objects/events in the adopted language, the anlage of future symbolic representations that give rise to the emergence of an inner language, the precursor of self-conscious thought. This is consciousness, however brief it may be functional in our quotidian existence. See http://drds.webhost4life.com/ID4-PlasticArt-com/index.htm In my opinion a machine cannot make axiological judgments, e.g., right/wrong, beyond what is pre-programmed, the will of the programmer not the free will of a self-conscious living human subject which btw, CAN be against biopsychosocial (BPS) self-interest if so freely willed. Dr.d
Reply | Report Abuse | Link to thisTo get a better understanding of the problem of artificial intelligence one should know that there is a difference between what is in a mind and what is real. Thinking in terms of mind space vs. real space helps to focus the issues involved.
Reply | Report Abuse | Link to thisPain is in mind space but not real space. Color is in mind space but not real space. Sound is in mind space but not real space. Light waves and sound waves are in real space.
Mind space is an enhanced representation of real space. Mind space correlates to real space so well that it seems that mind space is actually real space but its not.
People are conscious of mind space. Computers react to real space. Using algorithms, tables and random number generators the computer can mimic human actions, but in order to be conscious it needs to create its own mind space.
People can only experience their own mind space. They cannot observe another person's mind space (so far). So there is no way to confirm a computer's claim to mind space (or consciousness).
The analysis of a photograph to be correct or incorrect as a determinant of consciousness is invalid. "Consciousness" is by definition the awareness of self. The photograph test reveals nothing of this.
Reply | Report Abuse | Link to thisComputer software is much closer to accomplishing this than the authors realize. The tests proposed seem relatively simple, compared to what can currently be done on your home computer. The facial recognition software available to home users of, say, the Google Picasa product, can isolate a face on a photograph. To extend this to recognize other objects and then determine if the objects are appropriate together seems a small step. To match halves photograph parts separated with a black band seems simple when software like Photoshop can already eliminate telephone poles from photographs by extrapolating the background pixels.
This entire idea is an enormous red herring. The solution to the test proposed is a simple matter of processing data the same "un-sentient" way computers do today. The difference here is that the amount of data involved is much greater, and would require exhaustive libraries of all the possible objects that could be in a given picture, in addition to software to determine any environmental factors that might affect the picture, and libraries that contain all the possible "correct" arrangements of objects. This type of system would be able to beat this intelligence test. It would also be dumb as a rock.
Reply | Report Abuse | Link to thisThe bigger problem here, in my opinion, is that the two writers of this article are a neuroscientist and a psychologist, and neither is a computer programmer or involved in any of the attempts to develop an AI (though if I am wrong there, please correct me.) There are people working on the problems of Artificial Intelligence who do amazing work, and have created remarkable machines that impart at least the perception of intelligence, if not the perception of sentience. It speaks poorly of the writers of this article and of Scientific American for publishing it, that at no point do they mention discussing their idea with anyone actually involved in the field. I read this magazine for its coverage of science, not for someone's fanciful and ungrounded imaginings, however unique they may be.
An ape may be sentient or it may not be. What test might the ape take to clear this up?
Reply | Report Abuse | Link to thisIf we can't test an ape, then the computer test is invalid. It is specifically designed for a computer's output, not for the computer's mind.
It's not the questions we would ask of it. It's the questions it would ask of us, and who it would ask.
Reply | Report Abuse | Link to thisDoes a human have consciousness?
Reply | Report Abuse | Link to thisYour current consciousness is made up of absorbed data and experiences.
Does a Computer have consciousness?
A computer Absorbs data through a webcam, collects data through the microphone ect.
Does it think or not think?
The microphone will adjust the level of volume, based upon the level of input. Is that via choice? Did it decide to do that? Is it conscious because it changed its settings? A lot of people would say that a person is different from a computer, because they can think.
Well, defining thinking? I could say that the computer is thinking because its taking my voice and adjusting the volume accordingly. Its no different from a human thinking about having chicken tonight instead of pork. Or thinking about going to go pay there bill, or do whatever. Its ultimately just a more complex version of a computer. A program response. Functioning under the specific programs, one has accumulated from there upbringing and society.
If you ask someone a question and they give you a response, does that make them conscious? No. its still a function. Based on a programmed thought process.
When the computer adjusts its volume automatically. Who defined what level of sound would be ‘medium’. it was programmed with a opinion of someone’s idea, of what they thought ‘medium’ was.
We are all programmed with different opinions based on our individual life experiences. Ultimately our decision is made by someone else. Parent values/political values ect. Which were programmed into us, though out the years. No different from a computer being programmed.
Humans are not conscious. They Automate through life.
When your on the computer can you see outside of it? The wall behind it? can you feel the chair your sitting on? the temperature of the room, the breeze as it touches your skin? Are you aware of your body right now?
No, your completely absorbed in the reading these words. Your not aware of any of those things, unless its brought to your attention.
The human body is an organic machine.
Robots with equal intelligence to humans deserve equal rights.
cont.
Reply | Report Abuse | Link to thisBio-chemical organic reactions are patterned sequences the body learns to repeat. Such as seeing the face of a romantic partner, you feel overwhelmed with love because the brain releases a specific sequence of bio-chemicals every time you see that particular person. When that person goes away for a work trip for example you will withdraw from the emotional high your used to experiencing from having that person around and having them trigger that sequence. so the brain adapts by releasing a depressive combination of chemical reactions so that you (the combination of personality which make up your automated consciousness) go out of your way to resolve this problem by calling your partner or asking them to come home etc. So the body can have its 'love high' pattern repeated, which its become accustomed too.
Emotion is a programmed behavioral response for humans, just like it is for robots.
Sick today...flu... reading about consciousness. In my humble opinion it is a recognition that your view of an event is selected by you out of a multitude of possible views. You are 'aware' that there are other possible views but have 'chosen' one view. Further that one knows the chosen view makes one's reality or is totally meaningful but could be wrong and could also be totally meaningless. One person's complaint about someone else being on the muscle may be the other person's complaint the other is being defensive and can't take criticism. Freedom fighter or terrorist? It is when you are aware it is a choice and not some computational set of instructions that one is most conscious. Of course, having studied human beings for a long time, given their formative experiences (psych view), social situation (context) and known biological issues (bio aspects) one can predict what views other may hold. Despite that people can become aware they are making choices, examine themselves and change course. Therefore they are conscious... all others are zombies? Hmm...is this the fascination with Zombies? When a computer is aware of its own instruction set, can examine it, decide what view to take among a multitude of views and is aware that is only a view and not necessarily the 'right' view of events then I would deem it conscious. Conscious of itself as an actor in the world.
Reply | Report Abuse | Link to thisLet's make such computer first!
Reply | Report Abuse | Link to this