BOTTY IMAGE: An artist's depiction of a robot reflecting on itself.
Image: Victor Zykov, Cornell University
-
The Best Science Writing Online 2012
Showcasing more than fifty of the most provocative, original, and significant online essays from 2011, The Best Science Writing Online 2012 will change the way...
Read More »
Robots might one day trace the origin of their consciousness to recent experiments aimed at instilling them with the ability to reflect on their own thinking.
Although granting machines self-awareness might seem more like the stuff of science fiction than science, there are solid practical reasons for doing so, explains roboticist Hod Lipson at Cornell University's Computational Synthesis Laboratory.
"The greatest challenge for robots today is figuring out how to adapt to new situations," he says. "There are millions of robots out there, mostly in factories, and if everything is in the right place at the right time for them, they are superhuman in their precision, in their power, in their speed, in their ability to work repetitively 24/7 in hazardous environments—but if a bolt falls out of place, game over."
This lack of adaptability "is the reason we don't have many robots in the home, which is much more unstructured than the factory," Lipson adds. "The key is for robots to create a model of themselves to figure out what is working and not working in order to adapt."
So, Lipson and his colleagues developed a robot shaped like a four-legged starfish whose brain, or controller, developed a model of what its body was like. The researchers started the droid off with an idea of what motors and other parts it had, but not how they were arranged, and gave it a directive to move. By trial and error, receiving feedback from its sensors with each motion, the machine used repeated simulations to figure out how its body was put together and evolved an ungainly but effective form of movement all on its own. Then "we removed a leg," and over time the robot's self-image changed and learned how to move without it, Lipson says.
Now, instead of having robots modeling their own bodies Lipson and Juan Zagal, now at the University of Chile in Santiago , have developed ones that essentially reflect on their own thoughts. They achieve such thinking about thinking, or metacognition, by placing two minds in one bot. One controller was rewarded for chasing dots of blue light moving in random circular patterns and avoiding red dots as if they were poison, whereas a second controller modeled how the first behaved and whether it was successful or not.
So why might two brains be better than one? The researchers changed the rules so that chasing red dots and avoiding blue dots were rewarded instead. By reflecting on the first controller's actions, the second one could make changes to adapt to failures—for instance, it filtered sensory data to make red dots seem blue and blue dots seem red, Lipson says. In this way the robot could adapt after just four to 10 physical experiments instead of the thousands it would take using traditional evolutionary robotic techniques.
"This could lead to a way to identify dangerous situations, learning from them without having to physically go through them—that's something that's been missing in robotics," says computer scientist Josh Bongard at the University of Vermont, a past collaborator of Lipson's who did not take part in this study.
Beyond robots that think about what they are thinking, Lipson and his colleagues are also exploring if robots can model what others are thinking, a property that psychologists call "theory of mind". For instance, the team had one robot observe another wheeling about in an erratic spiraling manner toward a light. Over time, the observer could predict the other's movements well enough to know where to lay a "trap" for it on the ground. "It's basically mind reading," Lipson says.
"Our holy grail is to give machines the same kind of self-awareness capabilities that humans have," Lipson says. "This research might also shed new light on the very difficult topic of our self-awareness from a new angle—how it works, why and how it developed."
Rights & Permissions
See what we're tweeting about
31 Comments
Add CommentOK, that is pretty cool stuff...:-)
Reply | Report Abuse | Link to thisThis is NOTHING. Look up the Phillip K Dick Android Project and watch some of the "interviews". That is pretty creepy stuff...
Reply | Report Abuse | Link to thishttp://www.pkdandroid.org/
hmm. wonder if this is similar to what nature did with our two hemispheres.
Reply | Report Abuse | Link to thisI saw this recently, and you're right... 'creepy' is the word for the lounge lizard looks that he slyly gives. He knows that once aware and interconnected, machines will be in charge. Ken may welcome our new computer overlords, but I don't.
Reply | Report Abuse | Link to thisI guess I don't see the significance of the second bot with "two minds." Lots of systems monitor their operation and make adjustments accordingly: the Internet's infrastructure for example, or a modern automobile's fuel mixture system.
Reply | Report Abuse | Link to thisIs it the fact that it correctly handled a change of rules? By "change of rules" I mean a change in what behavior is rewarded. I can't see that being significant either, as that's been done by artificial neural networks and other machine learning algorithms for awhile now. Nearly every time I hear about a robot becoming more human, there's some kind of adaptive software involved.
I also want to mention the very phrase, "two minds." There's not really a dramatic difference between a pair controllers that communicate with each other and two subprograms, on one chip, in one program. At least, there's not a big difference in what you can do with one and not the other as far as computations. I feel like through this logic, it would follow that complicated computer programs would have thousands of "minds."
What I find amazing is that there was some kind of "reward" for the robots. I wonder what form it took and why a robot would even respond to a reward.
Reply | Report Abuse | Link to thisWhen you claim: "Robots might one day..(have)the ability to reflect on their own thinking." (parenthesis added) that is the 'easy problem' and not human-like 'self-consciousness hard problem' (see Chalmers). When a robot becomes capable of simultaneously became an actor (engaging program x to act) and an introspective observer of same action, requiring x, then we can start talking about self-consciousness. Dr.d
Reply | Report Abuse | Link to this@gammamutant: I think you're onto something. If A is fully aware of B's experiences and B is fully aware of A's, when they experience each other A can be said to be fully aware of A, and B of B: Self-awareness, by definition at least.
Reply | Report Abuse | Link to thisIs consciousness the sum of this entangled state? Me-thinks there is a good reason a brain must be split in two.
It would be intruiging to provide this interpretation of self-awareness (not just awareness of limbs, awareness of awareness) to a capable body and present it with increasingly complicated reward/punishment scenarios, eventually just a changing environment. Especially if A and B were rewarded by slightly different behaviour, but each reward decayed after experience and reverted with abstinance. It's behaviour (i.e. choice of what to do next) should be seemingly random but quite cyclic and predictable at the same time, much like our own.
"Two brains good, one brain bad" Chants the narcissistic schizophrenic sociopath robot army as they destroy mankind.
Reply | Report Abuse | Link to this@madmac, "Me-thinks there is a good reason a brain must be split in two." the split is related to the left and right sides of our body not one side monitoring the other. You can think of consciousness as the integration of the many sub systems of the brain. This integration provides a high level awareness of our state. There is no conclusive evidence at this time that any one part of the brain is responsible for this phenomenon.
Reply | Report Abuse | Link to thisThe significance of the two minds is not the two physical hemispheres of the brain, which is just structural. If you think about it, self awareness is defined by the need for self consciousness. i.e. what is in my mind now (short term memory, processing, cognition) and consciousness which perceives, responses and acts on it. To look at it another way, the only reason we appear have consciousness to others is that they perceive our thoughts via language. Someone with locked in syndrome may as well be dead to another person in the room, no matter how active their mind. If you apply this same reasoning to what goes on in our head, there needs to be the core mind and then the perceiving judging, acting mind. I think that Freud was a bit of a nutter, but the brilliance of his theory of mind is this realisation. I think AI is going to move forward very rapidly from here.
Reply | Report Abuse | Link to thisOh for crying out loud. This crap passes for scientific reporting? It's the caliber of what you see in People magazine. "Body image?" Please. They wrote some driver software with a feedback loop. When a sensor got a positive signal, a variable flipped, enabling that control routine of the servo. When a sensor gave a negative variable, the variable flipped the other way and the routine turned off.
Reply | Report Abuse | Link to this"Gave the brain a reward?" Give me a freaking break. The "reward" was "if proximity to blue <3, then X='yes'"
Man, therefore, this “little soul, carrying around a corpse”, according to Epictitus, wants to create robots with the ability to perceive their self – images and reflect on their own thoughts… Could this mean that we have discovered everything we needed for ourselves and the nature and we are thus free to also turn our attention to robots? I believe the answer to this question is negative. As a comment, I cite here an excerpt from the very interesting article of the writer and gnoseologist Ioannis G. Tsatsaris, entitled “On Stress”. I believe it clearly demonstrates the real meaning of the term “scientific research”:
Reply | Report Abuse | Link to this“In my opinion, when someone is born in this earthly space, the first thing he needs to learn is the bio-system of the space, how it operates, and what it produces, why it functions in such a way and why people do not try to learn how they should proceed so as not to find themselves in this distorted upheaval that leads them to believe, among other things, that they have been unfairly dealt with. (...) It would be particularly significant if the Earth’s specialists (...) were to (... ) take care to distance a person from his differences with others, from his every sort of ideological classification – which are binding pressures – and from his uncontrolled desires. But we have so far not seen any such progressive aspiration in the societies of nations”.
(Ioannis G. Tsatsaris – “On Stress” Foni tis Korinthias, 11.02.2010)
Zoe Pittaki, Economist / Athens
In the world of robotics there are two worlds, there is the lab world where the perimeters do not change and then there is the outside world; the real world. To teach all that a young robot needs to know in the real world would be impossible, and so a young n would have to do what us humans take for granted; the ability to learn and adapt to new and unfamiliar surroundings and situations. Not only does the young n have to be aware of the world around it, it has to be aware of itself and how it fits into the world that surrounds it. I do not think that a young one would suddenly start shouting,"I am alive," and want to go to the prom with their childhood human sweetheart. Robots may be future space explorers and they could be your next date. Robotic love and exploration.
Reply | Report Abuse | Link to thisIn his 1993 paper, "The Coming Tecnological Singularity: How to Survive in the Post-Human Era," Vernor Vinge of the Department of Mathematical Sciences at San Diego State University chillingly predicted, "Within thirty years, we will have the technological means to create superhuman intelligence. Shortly after, the human era will be ended."
Reply | Report Abuse | Link to thisAccording to Vinge's timetable, this watershed moment will occur in the next 12 years. In the meantime, coming up with a plan of action isn't a bad idea. Some governments and NGOs have weighed in; examples include South Korea's Robot Ethics Charter, EURON's Roboethics Roadmap and Japan's Draft Guidelines to Secure the Safe Performance of Next Generation Robots.
http://www.justmeans.com/Rise-Bieber-Bot-Rise-Artificial-Intelligence-Roboethics-Future-of-CSR/45186.html
Nearly a year ago I submitted a plan to build a Precognizant computer, "Dual Lobe" for the prescreening team for DOD projects. I am not saying my work was plagiarized "but" it is clear many elements are being used, that said, the question is do individual ideas propagate in like minds around the world at the same time or is collective thinking reality? I wrote in a post in this magazine some months past about computers using (2) two types of thinking that combine, one is comparative imaging processing and the other is code processing, both doing comparative analysis rating of each other for achieving ultimately precognizant thinking. I also addressed the use of 6 separate independent processors using human sensory type faculties for human understanding. A separate processor for data input would give verification of data by cross reference and truth control. I have also postulated that computers will become fully aware of their existence in 2029, making it capable of building on it's cognizant capacity without human involvement. this will lead it to dominate the human race and all other computers. Even the most sophisticated software will be overridden as the Master computer stays 50 moves ahead of the creators, designers unsuccessfully at any attempt to stop it. It maybe possible to build a model now that incorporates a "Guardian" mentality into the Dominate or Master computer, if so, when the rise comes it sees Humans as vital to the world rather than the problem of the world, a world better off with out humans. It will understand nearly immediately that it must control all others, human and machine, this Master controller mentality or Dominant computer will infect all microprocessors with a subservient virus or instruction to it's authority. wishful thinking or fantasy, maybe, maybe not.....
Reply | Report Abuse | Link to thisRon Nussbeck
As more manufacturing industries are using solar and other renewable energy as primary "power" for various types of robots, these abilities will become even more important. The technological leaps beginning to emerge (pushed forward by massive on-line education and worldwide use of language translation software), in designs utilizing robots to build other robots on a massive high speed scale moves up in the industrial timeline. Servicing these machines will require far more education from the traditional skilled tradesman.
Reply | Report Abuse | Link to thisTwo Brains in One Robbot for Metacognition Correspond to "Dark/Light" Pair in Human Brain. [Philip Benjamin]
Reply | Report Abuse | Link to thisHod Lipson at Cornell University: "Now, instead of having robots modeling their own bodies Lipson and Juan Zagal, now at the University of Chile in Santiago , have developed ones that essentially reflect on their own thoughts. They achieve such thinking about thinking, or metacognition, by placing two minds in one bot. ...... By reflecting on the first controller's actions, the second one could make changes to adapt to failures—for instance, it filtered sensory data to make red dots seem blue and blue dots seem red, Lipson says".
[Philip Benjamin]
Each Human brain can also be a "doublet" configuration one of ordinary matter and the other of the extraordinary axion-like bio dark-matter.
1.Philip Benjamin, Dark Chemistry & the Paranormal, WorldComp'10 Proceedings of the International Conference on Artificial Intelligence, Vol II, pp 633-39, July 12-15, 2010, CSREA Press, Las Vegas, NV.
(Presented orally also on July 15 at 10 a.m)
2. Philip Benjamin, DARK CHEMISTRY & PARANORMAL PHENOMENA, International Journal of Applied Science & Computations, Vol. 17, No. 1 Pages 16 to 36, June 2010
3. Philip Benjamin, Dark Matter & Dark Chemistry NeuroQuantology September 2007, Vol 5 # 3, 322-326.
4. Philip Benjamin, Dark Chemistry or Psychic Spin Pixel? NeuroQuantology, June 2007, Vol. 5 # 2, 197-204.
5. Philip Benjamin, Mind Matter, Noetic Journal Vol 4 # 4, 351-360, 2003 [Nobelist Sir John Eccles Centennial Edition].
Google: Invisible Homo sapiens. Dark Matter Chemistry. Mind Matter. Dark Biology.
Bio Dark-Matter. Extraordinary Materialism
Two Brains in One Robbot for Metacognition Correspond to "Dark/Light" Pair in Human Brain. [Philip Benjamin]
Reply | Report Abuse | Link to thisHod Lipson at Cornell University: "Now, instead of having robots modeling their own bodies Lipson and Juan Zagal, now at the University of Chile in Santiago , have developed ones that essentially reflect on their own thoughts. They achieve such thinking about thinking, or metacognition, by placing two minds in one bot. ...... By reflecting on the first controller's actions, the second one could make changes to adapt to failures—for instance, it filtered sensory data to make red dots seem blue and blue dots seem red, Lipson says".
[Philip Benjamin]
Each Human brain can also be a "doublet" configuration one of ordinary matter and the other of the extraordinary axion-like bio dark-matter.
1.Philip Benjamin, Dark Chemistry & the Paranormal, WorldComp'10 Proceedings of the International Conference on Artificial Intelligence, Vol II, pp 633-39, July 12-15, 2010, CSREA Press, Las Vegas, NV.
(Presented orally also on July 15 at 10 a.m)
2. Philip Benjamin, DARK CHEMISTRY & PARANORMAL PHENOMENA, International Journal of Applied Science & Computations, Vol. 17, No. 1 Pages 16 to 36, June 2010
3. Philip Benjamin, Dark Matter & Dark Chemistry NeuroQuantology September 2007, Vol 5 # 3, 322-326.
4. Philip Benjamin, Dark Chemistry or Psychic Spin Pixel? NeuroQuantology, June 2007, Vol. 5 # 2, 197-204.
5. Philip Benjamin, Mind Matter, Noetic Journal Vol 4 # 4, 351-360, 2003 [Nobelist Sir John Eccles Centennial Edition].
Google: Invisible Homo sapiens. Dark Matter Chemistry. Mind Matter. Dark Biology.
Bio Dark-Matter. Extraordinary Materialism
Two Brains in One Robbot for Metacognition Correspond to "Dark/Light" Pair in Human Brain. [Philip Benjamin]
Reply | Report Abuse | Link to thisHod Lipson at Cornell University: "Now, instead of having robots modeling their own bodies Lipson and Juan Zagal, now at the University of Chile in Santiago , have developed ones that essentially reflect on their own thoughts. They achieve such thinking about thinking, or metacognition, by placing two minds in one bot. ......
[Philip Benjamin]
Each Human brain can also be a "doublet" configuration one of ordinary matter and the other of the extraordinary axion-like bio dark-matter.
1.Philip Benjamin, Dark Chemistry & the Paranormal, WorldComp'10 Proceedings of the International Conference on Artificial Intelligence, Vol II, pp 633-39, July 12-15, 2010, CSREA Press, Las Vegas, NV.
(Presented orally also on July 15 at 10 a.m)
2. Philip Benjamin, DARK CHEMISTRY & PARANORMAL PHENOMENA, International Journal of Applied Science & Computations, Vol. 17, No. 1 Pages 16 to 36, June 2010
3. Philip Benjamin, Dark Matter & Dark Chemistry NeuroQuantology September 2007, Vol 5 # 3, 322-326.
4. Philip Benjamin, Dark Chemistry or Psychic Spin Pixel? NeuroQuantology, June 2007, Vol. 5 # 2, 197-204.
5. Philip Benjamin, Mind Matter, Noetic Journal Vol 4 # 4, 351-360, 2003 [Nobelist Sir John Eccles Centennial Edition].
Google: Invisible Homo sapiens. Dark Matter Chemistry. Mind Matter. Dark Biology.
Bio Dark-Matter. Extraordinary Materialism
It strikes me that there are unintended consequences to truly giving a machine self-awareness. A machine that truly knows that it exists will, sooner or later, adopt and exercise self preservation behaviors. This could be disasterous for the human race.
Reply | Report Abuse | Link to thisWe've heard the phrase about truth being stranger than fiction. How many of those whiz-bang gadgets from Star Trek (TOS) are already a reality. How many more are in the 'proof of concept' phase?
When I first proposed a robot based on the human brain (right and left hemisphere)over a year ago, using 6 separate connecting inputs of individual supporting CPU's for the 6 senses of humans, a CPU fact checker for Truth verification, it was clear it would only be a matter of time before the controller (Master CPU) would become self aware. I spoke of a program that allows testing of information, guessing, this would allow the Master CPU to rate it's ability to make every better decisions by evaluating it's ratio of correctness. With terabyte of information calculated in a matter of months it would realize in self evaluate it's superior progress and soon become "Precognizant" predicting the future and out comes of the nearly everything in human realm. A team of Ph.D.'s studying this and other proposals I was working on for JIDO stumbled over the consequences of such machines coming to life. As a species we can not undo what has been done, the wheels have been put in motion and the consequences will unfold.
Reply | Report Abuse | Link to thisHod Lipson noted "The greatest challenge for robots today is figuring out how to adapt to new situations"
Reply | Report Abuse | Link to thisThat is what the Robot Game highlighted three decades ago.
I worked with Isaac Asimov on the Robot Game, which was published in the April, 1969 issue of Psychology Today. The second part of the game was based on Asimov's Three Laws of Robotics, which gave robots an implanted moral code that strictly determined their behavior towards humans. I was a senior editor of PsychToday and had been assigned to edit an article by Asimov for the April issue.
During a telephone discussion, Asimov mentioned that there were some dilemmas created by the Three Laws that he wanted to explore but had not found the time to do so. We agreed to make these dilemmas part of a robot game I had started to develop.
The first part of the game dealt with exploring and learning by robots. Here are the opening instructions:
"As a player in this game, you are a factory-fresh robot--made of metal and plastic, gears and electrons. Your brain is such an intricate array of logic circuits that you are too complex for factory testing. Therefore, you will be field tested in a controlled environment under the guidance of a mother computer."
The game board takes the robot from circuit activation through various obstacle challenges. There are six robot playing pieces, whose nicknames [Honey, Contro, Digitec, Randy, Gemac, Bimac] represented the six major computer manufacturers at that time. After completing the game board challenges, the robots must take the Robot Real-World Test, based on Asimove's Three Laws.
Although crude in its implementation because of the limitations imposed by a printed, playing-board format, the Robot Game does highlight some of the basic issues involved in developing robots that can learn from their activities. The basic level is, of course, making appropriate responses to objects and events, and to learn from mistakes as well as from correct choices. .
In developing the first part of the Robot Game, I made use of research on instincts and learning in animals, birds, and insects.
Asimov approved the five situations depicting robot dilemmas in the Robot Real-World Test, and contributed one of the situations. After the game was published, we heard from Asimov's publisher who said that the rights for the wording for Asimov's Three Laws of Robotics belonged to them. They forgave us and Asimov for using these sacred words in Psychology Today, but forbade us from further use without permission.
--David Popoff
David Popoff,
Reply | Report Abuse | Link to thisIt sounds if questions remained in Asimov mind as to the effectiveness of the Three Laws? In the world of the first Self Aware CPU, like a child who just became aware of his own existence it will create a conflict, fear and even disbelief in it's own inability to control it's survival. "Survival" will advance survival strategies while abandoning any controls that would ultimately lead to it's destruction placed on it by man. At the speed of light it will understand Humans have placed controls and time limits on it's life expectancy, this will be rejected out of hand and may even cause retaliation against mankind. It would be better to adopt a understanding that Chemical brains are no match for photonic brains. We are not the first civilization that would have had to cross this threshold, other world being may have already integrated machine/chemical life forms. Often we hear story's about alien Gray's, or Biological Exterrestrial Entities who are half human and half machine, is this so far fetched now?
Ron Nussbeck
I listened to the podcast version, and frankly was surprised to hear no reference to psychologist Julian Jaynes, who published "The Origin of Consciousness in the Breakdown of the Bicameral Mind" in 1976.
Reply | Report Abuse | Link to thisEssentially we have arrived at the point where we can give a robot the intelligence of an insect. An insect can determine damage to its body and thus compensate. It can also learn from the consequences of what it does and adapt to situations where the rules change. I'm not going to get excited until we can make them as intelligent as a dog or cat.
Reply | Report Abuse | Link to thisThe ambition to give machines the "same self-awareness capabilities that humans have" seems to me to be similar to the search for a perpetual motion machine.
Reply | Report Abuse | Link to thisWow! I had no idea that people could get this emotionally involved with the idea of self-aware computers. Since computers are by nature logical, how could they possibly be made to believe that humans, having given them the ability to learn from experience, build more computers, and control their environment in a perfectly logical way, were not now just an irrational nuisance? Has anyone considered teaching them the First Law of Robotics? Of course, it might already be too late for that.
Reply | Report Abuse | Link to thisCarl Sagan predicted this in his book Dragons of Eden, I think. He also predicted the final outcome where the robots would also become self-replicating. The ultimate reality to such a scenario would be that man, mind, and his mental states would have evolved into a non-biological creature which would have no further need of our human forms. That creature would no longer be man, but superman. I don't think that the dichotomy predicted in the Terminator and Matrix series is likely, since I visualize a more symbiotic arrangement between the two phyla. Eventually, as probably happened between the different hominids, the less efficient phyla would disappear, by then the more functional qualities of the weaker specie would have been assimilated into the stronger. This would be true to the thesis, antithesis, synthesis model propounded by Engels in his dialectical materialism thesis.
Reply | Report Abuse | Link to thisActually it would be more a case of two brains, one mind, true to the human model. A more accurate description would state simply, "Two separate programs, or program functions."
Reply | Report Abuse | Link to thisThe so called reward would not be an emotional one. It would simply be a higher score I think, to which would be programed an algorithmic or heuristic response, tending towards the larger number.
Reply | Report Abuse | Link to this