When we last checked in with the Laboratory of Intelligent Systems in the Ecole Polytechnique Fédérale of Lausanne, Switzerland, their evolving robots had learned how deceive other robots about the location of a resource. Since then, their robots have continued to evolve, learning how to navigate a maze, beginning to cooperate and share, and even developing complex predator-prey interactions.
As before, the Swiss scientists placed within the robot's operating system both basic instructions, and some random variations that changed every generation in virtual mutations. After each trial, the code for the more successful robots got passed on to the next generation, while the code for the less successful robots got bred out.
This time, however, the researchers designed a whole new menagerie of robots, including a set of hunter robots that pursue prey-bots, maze-running robots, and robots designed to deposit a token in a given area.
For the first experiment, the scientists created two sets of bots: predator bots with better eyesight, and prey bots with more speed. Initially, the predator was only programmed to find the prey and then drive towards it, while the prey was only programmed to move away when it detected the predator. At first, the robots just bounced towards and away from each other randomly. But over 125 generations, the hunter-bots learned to approach the prey from blind spots and to hide against the walls in wait, while the prey-bots learned to stay away from the walls and retreat with its sensors facing the hunter-bots, so it could keep the danger in sight.
In the maze experiment, robots with six sensors on one side and two sensors on the other started with the basic programming of running the maze, and reproducing less if the sensors were trigger by a bump into the wall. After less than 100 generations, the robots had not only evolved the ability to navigate the maze without any wall collisions, but even learned to have the side with more sensors face the direction of travel.
With the final experiment, the scientists created robots that got points for placing tokens in a marked area. The more points, the more offspring. The catch was two types of tokens: one small enough for an individual to push, but worth fewer points, and a bigger token requiring two robots to move, but worth more points. Not only did the robots evolve to help each other, but like in nature, they evolved to only help those robots from the same code lineage, a trait called "kin selection" in biology.
Most amazingly, the code for the robots in all the experiments was amazingly short. In fact, for the token-moving experiment, the robots only had the programing equivalent of 15 neurons. By coaxing such complex behavior out of limited programming, the the Laboratory of Intelligent Systems team proved, once again, that some of nature's most complex behaviors are emergent phenomena that grow out of very simple instructions.
This is fun stuff and will be a great deal of help in evolving out true AI.
We can evolve out instinct programs and then implant them along with basic senses programs and routines then bring them all together into a fully sensor rigged unit and raise it.
This is the most likely path to true AI everything else will be too predictable.
it seems lik if humans went back 200,000 years the simple instructions were have babies eat and stay warm
Good point Stuntman, seems like some things never change :)
I like the "strength of connections" method of determining behaviour, good work.
Finally we are starting to talk about the fact that AI is not a fancy human shell, nice motor movements and canned TTS but desperately needs software if we are ever to achieve intelligence in robotics.
I was actually touched to learn that the token-bots developed such cooperative behavior - particularly with just 15 neurons' worth of complexity - and that they were able to recognize and work with their "kin".
If such simple robots can learn the basics of friendship and family attachments, maybe our relationship with future AIs could be more of a partnership. It would certainly be a better alternative than the predator-and-prey conflict you see in a lot of science fiction (e.g. Skynet, Cylons, etc.).
A nice sentiment Q42. But then you must define (or redefine kinship) because these robots don't understand what they're doing. Self-awarness is not part of these pseduo-nerons. The only reason these robots act kindly towards kin is because the random coding that caused that is beneficial, or is part of a benficial piece of coding that has lived via virtual evolution.
For example, with these same robots, the company made a thing where the robots try to find this "food" and avoid this "poison" (two devices they could detect). Some robots developed a form of "lying" where they were originally programmed to alert everyone else when they found the food, but instead they did not. This devloped because the "food" was scarce (only so many can huddle around it) and the program got points for time spent in proximity of the "food". So deception evolved for the same reason that kinship evolved in this scenario. It was personally beneficial.
The next proper step, however, is to use a few code genomes and make groups. Evolution is then determined by group success. It is there you will see the unbeneficial traits like mercy and fairness arise, because the colony with robots like that will very likely prosper. Mutual beneficiency only comes from evolution as a group, rarley as an individual.
let us not forget the most important thing about computers or robots that evolve. It just goes to show that when something is designed to evolve, it does. If this doesn't prove that life evolved without a designer, I don't know what will!
Think about what you said in your comment... <<< "let us not forget the most important thing about computers or robots that evolve. It just goes to show that WHEN SOMETHING IS DESIGNED TO EVOLVE, it does. If this doesn't prove that life evolved WITHOUT A DESIGNER, I don't know what will!" >>>
<<< "something is DESIGNED... without a DESIGNER..." >>>
How does that make sense?
What geebob said is entirely self-consistent and correct. *These robots* are clearly designed (and designed to evolve). From that starting point, they evolve behaviours that we see in life -- such as cooperation between kin, deception, learning, self-preservation, etc. (in the same understanding-free senses that these behaviours occur in, say, insects), without the need for any *further* design input.
This tells us that the main assumptions behind "intelligent design" or creationism -- that an "intelligent designer" is *necessary* in order to explain why organisms have these behaviours -- are simply wrong. To demonstrate that the initial conditions for evolution can arise purely by physical processes is a separate problem (one on which spectacular progress is being made, as it happens). But that's not necessary to show that life "evolved without a designer".
You people are so desperate for ANY evidence (no matter how circumstantial) to support Darwinian Evolution you will grasp at anything that (supposedly) supports your view point.
This "evolution" of behavior by robots has taken many many years of AI research by some of the most intelligent human beings (in the field of computer science) we could produce. By just looking at the evolved behaviors you completely ignore the complexity of designing the silicon "neurons" and the programming code required to choose what has been successful and what has not been successful. How many lines of programming code is necessary for these robots to "evolve"? How many lines of code are necessary for the robot to choose how to move and which way to turn?
As a programmer I will guarantee you these robots require hundreds, if not thousands, of lines of complex decision making code before than can even begin to evolve behaviors.
No matter how successful any future neural network becomes at "evolving" behaviors, it does not in any way support the theory of Darwinian evolution or disprove in any way the theory of Intelligent Design.
Here is a challenge for anyone who believes in Darwins' theory. Gather the greatest scientific minds in the fields of biology and chemistry the world has to offer. Let them choose any species of mammal they want to work with, and let them conceive of and design the DNA for a new internal organ that has NEVER been seen in nature before. The new organ must confer some advantage for the species survival. Then have them incorporate the DNA for the new organ into the DNA of the mammal. Hey, if a world completely void of intelligence can design and create complex organs for every known species surely the brilliant minds of our scientists can do it. Umm...Good luck with that.
"Here is a challenge for anyone who believes in Darwins' theory. Gather the greatest scientific minds in the fields of biology and chemistry the world has to offer. Let them choose any species of mammal they want to work with, and let them conceive of and design the DNA for a new internal organ that has NEVER been seen in nature before. The new organ must confer some advantage for the species survival. Then have them incorporate the DNA for the new organ into the DNA of the mammal. Hey, if a world completely void of intelligence can design and create complex organs for every known species surely the brilliant minds of our scientists can do it. Umm...Good luck with that."
how would this prove anything? you need to read the book Your inner fish. If you want to learn about the evolutionary process and cell manipulation. The book explains each cellular change to create new species as well as how to reverse engineer the body to its basic parts.
RealCoolDude said: "how would this prove anything? you need to read the book Your inner fish. If you want to learn about the evolutionary process and cell manipulation. The book explains each cellular change to create new species as well as how to reverse engineer the body to its basic parts."
You are missing the most important point. So what if scientists can take something that already exists and reverse engineer it. The main point I was making is to conceive of then design something new, something never before observed in nature, design its function and how it achieves what it is supposed to do. Then create the DNA sequence that could be incorporated into the mammal and passed on from generation to generation. This is supposedly what evolution has done completely without any intelligence whatsoever. Let's see if mankind can do it.
theconspiracy03, what you are missing is that evolution is a trial and error system, and the combined lifetimes of the world's smartest scientists is abysmally short compared to the time it takes for a new organ to evolve in nature. The human understanding of genes isn't advanced enough for much more than trial and error at this point, however if you gave a monkey a flip to switch to randomly change nucleotides and 5 million years he could very well create a new organ assuming that the undesirable mutations died off at a similiar rate to that of nature.
I guess "Higher Intelligence" isn't as "High" as we like to believe.