....
....
Hello reader.
This is a copy of "What do robots think about?" by the famous French author Jean-Pierre Petit. Originally written in 1982, and translated into russian in 1987, this book remains highly relevant in the field of robotics (the first chapter could be the inspiration for the Roomba). Packaged in the form of comics, these concepts were a lot of fun for me as a child, and I hope you will enjoy them too.
This is a literal translation, and some ideas are sure to have been lost, so for another translation, as well as many ther books, and languages, visit the author's site.
FRONT COVER:
WHAT DO ROBOTS THINK ABOUT?
1) Look at that, people!...
2) I wonder what they're going to do?
Inside cover 1
Pelican) We expect that robots don't think...
Inside cover 2
Snail) Robots can do anything, but what do they think about it?
(publishing pages translation omitted)
PROLOGUE (page 7)
Pe) Anselmi, what are you constructing?
Sn) Is it a ship?
An) You'll see soon. This is.. the... hull
Dr. Bird) This is no ship!
Sn) Excellent! It's a steam-powered pelican!
Pe) Hwa-ha?
page 8)
Sn) Yep, that's the thing! It even waddles like a pelican
Pe) Oh, sure, Terezie, when I can be made the butt of the joke, you're ready to oblidge!
- Ah, look at that...
Dr) A hovercraft snail!
< Klok, klok, klok >
Sophie) Oh, great, look at that, more work!
page 9)
An) So what's wrong with that?
So) Once again, you've turned our living room into a laboratory. Everythig's turned upside-down!
Dr) So long ...
An) Stupid device
- all finished
- It's so hot around here
- Hmmmm
Pe) Anselmi's completely broke his vaccum cleaner.
Dr) Yes and no, he's optimizing it
page 10)
PROGRAMMABLE AUTOMATA
An) Now I have a programmable vaccume.
- Look; I can turn it on and control it with the help of these levers...
Dr) So what's the difference? Instead of controlling it with your hands, you now have to control it with these levers?
An) Left! Right !
- When the work is done, I drive it to a device that recharges its battery.
Dr) So, what's so interesting about this?
An) I position my TURTLE by the bin, where she deposits the dust.
page 11)
An) When the work is done, I drive it to a device that recharges its battery.
Dr) So what's so interesting about this?
An) You'll see in a second. I'm going to press this lever,
- and my Turtle can redo the same path, going around the furniture!
page 12)
An) And then the dust empties where it's supposed to!
Dr) And then it returns to recharge ...
And) Let me reveal the mystery...
(Clockwise, starting on the very top left)
- levers for mannual operation
- contacts for recharging
- battery
- punch tape
- turbine of the vaccuum
- freely moving front wheel
- independently controlled back wheels
- electromagnet for opening the dust box
An) When I turn on the Turtle, this tape begins to unwind with a constant speed. it has commands written on it as perforations.
- In this way, the entire task is written in the MEMORY of the Turtle.
page 13)
An) Next, we'll attach the Turtle to a charging station that rewinds the tape
Dr) This means that you can write a program and make the Turtle understand and execute commands? But where did you find the tape?
An) I took it out of an old piano
- Cleanup in the kitchen!
- ?
- Sophie, watch out!
page 14)
An) Now we can go out, and the turtle can take care of cleaning the house.
So) Good job Lanturlu
(Three hours later..)
So: My livingroom! What a disaster!
An) Oh-oh. I'm about to get it...
(Analysis of the situation allows us to deduce what happened.
- the Turtle encountered the cat's tail, but this was not forseen in its programming.
- She immediately vaccumed up the tail.
Cat) Meeeeoooow!
page 15)
- In a terrible panic, the cat dragged the robo-turtle on its tail around the room - destroying all in his wake.
Rat) Help!
So) The acid from the battery spilled on the rug and ate through it. The Turtle even sucked up my tights.
An) Ok, ok, I'll start cleaning it up.
page 16)
SYSTEMS WITH INPUTS AND OUTPUTS
So) The cat left a note, "Farewell, I cannot remain another moinute in this madhouse. Do not expect me to catch mice for you anymore."
Su) Without a warning!
An) My Turtle, that I thought was pretty clever, turns out to be very dumb.
So) Why would it be smart? it doesn't know what's going on around it! It's imperative that we supply it wish sensors.
An) You are right, we need to make my turtle see. This device is sensitive to INFRARED, i.e. heat radiation.
Sn) And in what way will it react to heat?
- heat radiation
page 17)
An) With a couple of hundred of these elements, I have created a basic "eye," similar to the infrared eyes of a snake.
An) Our tile floor is cold. For the Turtle, this will be baseline temperature. In this way, it will be percieved as black.
An) Our walls, which have paper wallpaper, are slightly warmer, and so they'll be percieved as gray.
Sn) The field of vision is divided into cells to ease perception.
Page 18)
- The field of vision of the Turtle is a cone.
Dr) The Turtle can tell the angle between an object and its optic axis.
Dr) Each infrared element corresponds to a pixel MxM sized 1(degree)x1(degree), which determines the resolution power of the optical system; the image is RASTERIZED that is, turned into white, black, and gray rectangles.
An) I've junked all the old programs, and I plan to give the Turtle a simple reflex that will allow it to avoid obstacles. If one of the photosensitive elements in the turtle's eye recieves radiation whose intensity exeeds the value of the floor; and if this source of heat is below the ocular axis, and the angle phi is less than phi(min), the Turtle will take a 90 degree turn to the right.
Box) When the intensity of the heat radiation is above the threshold, and angle psi is less than dthe horizon value of psi(min),
Dr) That means the signal is in the shaded part of the circle
Tr) So I turn
page 19)
So) In this manner, the Turtle represents a SYSTEM WITH INPUTS AND OUTPUTS
-"BLACK BOX"
Sn) What's in this "black box"?
Dr) A PROGRAM that analyzes the input signal (Information processing net) and on that basis, determines the future actions, for example turning and moving right.
An) Now, my turtle will suck up dirt and won't run into anything. She will avoid walls, and also furniture - which is even darker than the floor.
An) It's all in order. The problem is solved. Before, it was all off, but now I'm sure I've got it straight.
Pe) Anselmi!!!
An) What?
Pe) The Turtle has sucked up Terezin!
An) Seriously? Byt why?
page 20)
Dr) Terezin is cold blooded and also her body is positioned ner the floor. That is why she had the same darkness as the floor and was invisible to the turtle.
Sn) Careful!
An) I connected a candle to her back. Now she'll be safe.
So) Look, when the wall is at an acute angle, as here, the turtle has to turn twice. That's kind of awkward.
An) Yeah, it would be better to turn left here.
SIGNAL ANALYSIS
An) How should I do this?
- I know!
Box) OLD PROGRAM
If a warm object appears in region A, turn right.
Box) NEW PROGRAM
If a warm object appears in region A, turn right.
If it appears in region B, turn left.
page 21)
Sn) Ok, but what if the turtle comes to the wall at a right angle?
Pe) Anselmi, she'll burn up!
An) Dumb animal!
Dr) The solution is this: enter into the turning algorithm - if the signal demands a simultaneous turn to the right and left, turn 90 degrees right.
An) Sophie! I've thought of how to improve the interpretation of the signals recieved by the eyes of the Turtle. For example, if this object comes into its field of vision, she only has to count up the number of elements comrising the object in field A(left) and in field B(right).
So) You also have to forsee the Turtle being able to go between objects.
page 22)
An) For that, It is sufficient to reduce zones A and B in the display of the turtle.
REACTION TIME
An) A device is interesting if it works fast. I'm going to maximize the speed of the turtle so that it does the work fastest.
- Oh - no! Not again!
page 23)
So) Anselmi - you can't expect the system to react to inputs immediately. Every self-controlled system has REACTION TIME.
- Because of reaction time, you can't catch the bank-note when I let it go.
An) I missed it again!
Rat) It's imperative to be devilishly fast.
Box) Attention! This Input/Output system has a short response time.
An) Cleanup takes a lot of energy. My Turtle is completely spent.
Sn) The turtle should automatically return to recharge.
Dr) Sure, but how?
page 24)
FEEDBACK SYSTEMS
An) When the charge in the battery decreases, so does the voltage. It's easy to include, in the program, something like this: IF the voltage is below a certain level, THEN return to recharge. There's still a question - how does the turtle find the charges?
- Between projects, I made a device that can recharge itself using solar rays.
Sn) We can call it a POWER PLANT
An) This wire can be heated to direct the turtle, but how?
So) You have to include FEEDBACK into the Turtle.
page 25)
So) The wire should be be seen as an obstacle. We should hang it up as high as possible.
Pe) And so, the potential will show up in the top segment of the display.
Sn) It's all so simple! If the charge of the batteries falls below a certain voltage, th eTurtle will start looking for a heat source. There are two possibilities: Either the coil is in the field of vision and she needs to drive to it, or it isn't and the turtle should turn until it is.
So) As soon as the coil is visible, the turtle will drive to it with the following program:
Box) Correlate the direction of movement proportionally to the size of the angle and in the direction of the angle. This is called LINEAR FEEDBACK.
page 26)
- With a change of course, the angle psi also changes. This is how ANGULAR direction works.
Tu) Turn just - but to the right.
Dr) This is how the turtle directs itself to the target.
ORDER OF FEEDBACK SYSTEMS
So) Hey Anselmi: Let me present you with a game. On my signal, you have to put the point of your pencil on the point of mine.
An) Aha, you want to say that I present a feedback system - "black box"
page 27)
So) Some sort of "TERMINATOR"
An) I direct my pencil with the observable angular difference.
So) Come on, faster!
An) What's going on here? I can position my pencil but only after a couple of correlations (attempts).
So: This happens because of INERTIA
Dr) In a FIRST ORDER SYSTEM, the direction command works directly on speed without regard for inertia. In this situation, there are no vibrations.
In a SECOND ORDER SYSTEM the direction commands (expressed as a force) act on acceleration. Thanks to inertia, this can lead to oscillations.
page 28)
Dr) A person, thus, is a 2nd order system. The flush tank in a toilet is a good example of a first order system. Speed of the water rising is proportional to the height (e) needed to rise.
Pe) In truth, this is so, the height of the water tank does not vibrate.
Dr) But due to the physical nature of things, INERTIA appears all the time. If they're pushed a little, a first order system can turn into a second order one.
page 29)
Sn) Sophie, Leon said that I move so slow that I'll never feel inertia, and so I must be a first order system.
SELF-REGULATING SYSTEMS
Dr) Let's return to our turtle
Dr) Anselmi made a clever system. The Power Plant creates a difference in potential between two circular wires. The copper wheels of the turtle bridge the contact. As soon as the front wheel hits the anode, and the back, the cathode, the turtle stops and charges. When charged, turns around and returns to work. When it runs out again, the heated coil directs the turtle to the charging station.
page 30)
So) You also stop eating when the pressure on the walls of your stomach exceede a certain point.
An) Who me..?
Sn) Yeah, you. The stomach works exactly the same way as the toilet bucket.
Dr) No, the toilet bucket acts like a bucket.
Pe) Ugh, stop being so vulgar.
So) These are examples of systems that seek to stay within their parameters by regulating between a minimum and maximum value.
An) I propose that when I eat, I seek to regulate the amount of sugar, salt, etc... in a balance between some maximum and minimum value.
page 31)
An) So then why should I make something similar in a machine?
So) I'd say the opposite - this machine is similar to a person.
- Everything that's made by people; techniques and technology - are ways of recreating or mimiking natural processes.
- After humans learned how to copy the forms of nature, they started copying the methods.
Box) An Input/Output system with a quick response
page 32)
So) A living organism - it's a fantastic machine; it's self-regulating and homeostatic; it maintains the right amounts of water salt, blood, muscles, and also its own form.
- Cells also stop their growth when they come into contact
Dr) Any noticeable deviation of values from the standard results in the destruction of tissues.
Sn) Any disruption of the cellular cover results in increased cell division on the sides of the wound.
Dr) The division of cells stops as soon as contact between them is restored.
page 33)
Dr) If this process of healing stops acting too slowly, inflamation will occur.
Pe) What are you doing?
An) Sophie asked me to rid the house of mice. I have to catch them all.
I'll make the turtle an arm that coomerates with its eye.
Dr) Mice are warm. 42 degrees c
An) This is necessary for determining the baseline. The floor is "black", the walls are "grey", the mice are "white". We just have to edit the program to chase any object that's hot enough (higher than the constant C which is written on the magnetic tape)
page 34)
An)First it fids a mouse
M) What kind of thing is this?
An) Then the turtle aims from above.
Dr) Then it nears until the mouse is in the zone of capture.
M) Let me go! What is this!
An) There we go.
page 35)
So) Anselmi!
An) What?
Sn) Is this about me?
So) Can you make this thing let go of the cup, or at least what's left of it?
An) Of course, it can't tell a mouse from a cup of warm coffee
Snake) The devil!
Dr) The same thing can happen to a snake when she goes hunting at night
An) My Turtle can get confused by a nearby target and a far one. The intensity of the radiation varies with the inverse square of the distance to it.
Sn) The turtle is nearsighted!
page 36)
So) Don't forget that when winter comes, your Turtle will be totally blind.
An) Why?
So) Because we have a heated floor.
An) The situation is hopeless.
Dr) Maybe you can make it hybernate
Sn) We need to teach the Turtle to RECOGNIZE SHAPES of objects to tell them from one another.
Pe) While you're at it, why not teach it to read?
page 37)
PATTERN RECOGNITION
An) This video camera with a high resolution will fit us a lot better than the infrared eye.
So) The display of any object that's in front of the camera consists of a collection of small squares with x and y coordinates.
Dr) For some shape to be recognized, it needs to be known
Dr) Let's start by teaching the Turtle to MEMORIZE SHAPES
page 38)
Dr) For example, letters of the alphabet
Pe) It's clear to me what needs to be done. The letters need to be shown to the machine one after another, and it needs to compare them to images that it knows.
Dr) The signal being analyzed should be overlayed a known signal - one that is stored as a negative.
An) That way, if the signals correspond, the field will be consistently grey.
Page 39)
Dr) Actually, each image is represented by a collection of ones and zeroes that are organized by their x and y locations. The integrated computer will save all the differences and coincidences. Coincidences - 4 cases. Number of cells in signal - 7. Ratio of coincidences to cells - 4/7
Pe) Sure, but there's still a lot of problems. The turtle can't recognize a letter if it's tilted or at the wrong distance.
Pe) Are you relying on good luck?
An) The devil! What should I do?
So) I know!
page 40)
So) I find the center of gravity of each symbol
So) Before recording the symbol in the memory of the machine, I'm going to have to record its center as well.
So) That way, whenever a symbol enters the field of vision
S0) I can immediately make it compute the center of the part that it sees, G1, and then align the center of the camera with the center of the image.
page 41)
So) When this is done, the machine can compute the center of G2 the new image that it now sees, and then allign with it.
So) In this way, it can hone in on the center of the object.
Pe) But we're still left with the problem of the size of the object.
- The integrated computer can make N copies of the image, increased or reduced.
So) And then compare these images to the saved ones in the memory.
Tu) Oh yeah, it's a P!
page 42)
An) Wait a second! We can get rid of these systematic comparisons of images. From afar, each letter looks like an indistinct object. Each of these images has a center and a CHARACTERISTIC DIAMETER
Sn) But how are you going to determine this diameter?
An) I'm going to take each of the darkened squares M and find the distance from them to the center G. Next, I'm going to add all the lengths GM and divide by their number. This average of lengths gives me a radius R and then D becomes 2R.
An) That way we can assign a characteristic diameter to each letter and symbol.
An) Instead of memorizing all the symbols, I'm going to give them all a center of Xg = 0 and Yg=0 - and the same Diameter, for example D = 1.
Sn) In other words, you center and focus the shape.
page 43)
So) In this way, we've established, that the Turtle can automatically find the center of the image, then by averaging points find it's diameter D. If D is different from 1, the Turtle's computer can shrink or grow the image until the Diameter is 1.
Sn) Now all that's left is to go through all the images in memory and look for a match.
page 44)
Sn) But what if we include the possibility that the image can be tilted with respect to the horizon? Then we'll have to turn each image from 0 to 360 degrees.
Pe) If people worked that way, each hard job would have to be done at each angle as well! This would take much more time.
So) If we try to do this with only one MICROPROCESSOR, like the Turtle, then you're right. But you can't say the same about a system of thousands of microprocessors, all working IN PARALLEL.
-...No, that's not it
-...No, not it
- Hey guys! I found it!
Sn) The time for INFORMATION PROCESSING is a lot smaller.
page 45)
Dr) A person can recognize letters at reading speed. Your brain can decode information that's been sent by your eyes as fast as 10,000 microprocessors working in parallel.
Sn) And really, it's a wonder when a person memorizes an image, he's ONE.
Dr) Boggles the mind
So) General pattern recognition is a much more difficult problem. We've only shown a bare outline of the problem.
Pe) Is this the first Turtle that can read?
Pe) Would you look at that...
page 46)
Dr) If the turtle can recognize letters, then it can recognize groups of letters, words, and whole phrases.
Box) DANGER
An) We're about to see a very interesting experiment.
Sn) It's snowing
Box) This is the image of a letter that the Turtle can see at a certain moment.
page 47)
NOISE
Dr) Each letter is disturbed by BACKGROUND NOISE. First idea: There's no 100% correlation to any letter.
An) Despite the noise, I can still tell these letters apart. Even though they're no more than 75% complete.
Sn) But here, we can't tell anything.
So) But we can superimpose many images of the object on the same location.
Dr) Since you have two eyes, at any moment, you're looking at a superimposition of two images.
page 48)
IMAGE PROCESSING
So) A combination of N images can improve the signal, making it more clear.
Dr) We can, for example, put them on top of each other and average.
So) There are many more ways to combine images with mathematical methods.
An) When I block the signal with my fingers, I can't read it.
An) But when I move my hand back and forth, I can!
page 49)
Pe) If I understood you correctly, you want to teach your Turtle to recognize shapes and read with noisy conditions. But object recognition is dependent on finding it's center and diameter. The BACKGROUND NOISE makes that problematic...
So) I have an idea how to "clean up" the image. We can examine each pixel and and eliminate each one that doesn't have at least two neigbors.
So) This will let is get rid of a large number of isolated points of noise.
Dr) Then you can invert the image and do it again.
Sn) Here's the result of the second "clean up"
page 50)
Pe) Fine, so let's say the Turtle can read in any weather? Why do we need that?
Dr) Imagine, Leon, that the internal systems of the camera become similar to a human's.
Sn) The number of pixels at the center of the view will become much greater than on the edges.
WHEN READING TEXT the axis of sight moves
around the symbols erratically only a few parts of the image are shown clearly,
but the mind reconstructs enough of the text to find a MOST LIKELY sentence.
So) This lets us read very rapidly.
page 51)
So) In truth, GATHERING all the information, going from word to word - letter after letter - element after element - is unending work
Dr) To tell symbols apart, it's enough to see "key" forms of the image
Pe) I see. That information is extra
An) In most cases, it's enough to look at the beginning, but not the end of the word.
An) And what you can say about reading, you can say about any visual perception.
An) When a person thinks that there's something wrong with his image, his eye looks over the same area many times quickly.
page 52)
PERCEPTION
Dr) Our sensory organs are always getting information about the surrounding world.
So) But from all the information we recieve, we only process a MINIMUM required to tell apart shapes, sounds, etc.
So) These images are then compared to the enormous bank of images and sounds we have in our memory...
- memory written in the sand
Sn) Which are constantly updated
Dr) It's determined that if you deprive somebody of all information from the outside world, he will soon become UNNORMAL.
page 53)
Sn) After a few days of total sensory deprivation, he can't complete even basic motor movements, like for example, holding a glass. This gives the impression that our impressions of the outside world need constant reinforcement.
Dr) We are in a state of constant learning.
Pe) Alright, but let's go back to our Turtle. She can quickly and constantly recognize different shapes. But WHAT NEXT?
Pe) She's still dumb.
page 54)
Pe) She can only do what's written in her memory.
Signs) Enter, Stop, Exit, Food, Attention
So) We can add a generator of random signals to make the actions a little different.
Sn) But it will at some point have to make desicions.
Dr) All living things have systems for making desicions.
Pe) I have to admit, I sometimes have to do things... that I don't understand...
page 55)
INTELLIGENCE AND STUPIDITY
Pe) This is all illusion. Every machine is and will always be completely stupid.
so) Leon, what is intelligence?
Pe) Intelligence... I have to quickly find a definition! I have to admit I'm not a person...
An) Often, we act like a real mechanism, as if programmed...
Sn) A person that can only march to a beat doesn't need a brain. A spinal chord is quite sufficient. *
* Albert Einstein said this
page 56)
An) Intelligence - this is a thing that floats on top of an ocean of reflexes and goals.
So) What we've done so far is try to mimic the behavior of animals on different levels of evolution.
So) But even though no newborn can speak or dance, still from the moment of birth, he has a genetically PREPROGRAMMED INSTRUCTIONS of behavior, i.e. INSTINCTS.
Dr) The teach, in part, to have an ambition to explore and to learn.
An) I'd like to teach my turtle that way. So that it would be ameable to learning.
page 57
METHOD OF TRIAL AND ERROR
Tu) Interesting, this is a word that's unknown to me. But it looks like the arrow is showing a direction.
Tu) I'll enter this word into memory
Box) STAIRS
Tu) Let's check it out...
page 58)
Tu) Now I know that I don't like stairs.
- The turtle can carry out experiments and then use the results to its own benefit.
Tu) At sixty volts I can charge very fast, but I overheat.
So) It would be nice to know what we can learn with this method of learning.
page 59)
Dr) A machine can store any type of data in its memory.
Reels) Chess rules / Chess games 1890-1912 / Chess games 1913 - 1922
Sn) First, you need access to a DATA BANK
So) Then, after using its sense organs
Dr) These data can be ANALYZED and connections can be found between them.
Tu) I've noticed that the chance of rain increases when dark masses gather in front of the sun.
page 60)
Tu) And rain is bad for my circuits.
Sn) These continual experiments cause the Turtle to keep changing its STRATEGY.
Sn) Ah, what a wonderful weather we have today.
Tu) Hmmm, let me think, last time I was in this situation, a rook move was a bad idea.
Tu) Using this new information, and the game between Alekhine and Morphy in 1924, the pawn looks like it can determine the game.
Tu) Let's try that operation
Tu) Let's see...
page 61)
So) O - ho - ho
Sn) And so, where does intelect begin and stupidity end?
TURING TEST
- a "mathematician" developed a test for intellect
Sn) We shouldn't forget that after 1981 the chess champion has become the computer.
An) Hello? Mademuselle, you say that there's a spot on the 22:30 train?
Telephone) Yes Monsur, I'll reserve you a seat.
An) Did you know you have a charming voice?
Telephone) Please Monsur, be serious.
An) Are you free some night this week?
Telephone) Monsur, this is impossible
An) But why is that?
page 62)
An) Hmm, let me try a test.
An) What is your opinion on waffles?
Telephone) Waffles? What county is that in?
An) Again, it's one of those damn machines
Telephone) Hello! Monsur? Your reservation!
Box) A machine can be considered intelligent if its behavior cannot be distinguished from a person's. - Turing
Pe) Thank god, there's still a long way to go.
Pelican 2) Artificial intelligence? Ptui!
page 63)
Pelican 2) All of this is just tricks. Nobody is going to convice me that one day, a computer can compare intelligences with...
Sn) A bird-brain?
Pe) O-o-o watch it!
Sn) he-he-he
Sn) We are in danger of not noticing when one beautiful day, machines get intelligence, and we will not be ready for it.
Hand) Terezia!
Dr) What bunk!
page 64)
ARTIFICIAL INTELLIGENCE
An) How to get out?
An) You just mentioned, Sophie, that we've started making machines that simulate our living things.
Pe) That's not to be taken literally. Even a car resembles a horse.
So) Yes, but both a car and a horse take in chemical energy, and convert it into mechanical energy of movement, and so the are FUNCTIONALLY engines.
Sn) Yes, that's true, a machine copies and complements nature.
page 65)
-With the help of CYBERNATICS and INFORMATION SCIENCE, humans have imitated the functions of REGULATION and PROGRAMMING living things, imitating instincts.
So) For scrap metal!
Tu) Run for it!
Sn) Instinctive response
Tu) It's getting cold. I'll turn on the heating.
Pe) It's hard to stop on this good path
Dr) People have also been able to give machines:
Dr) Ability to LEARN
Sign) Bridge out. Detour 500m
- Ability to ASSOCIATE
Tu) If all greeks are liars, and if Epimenides is greek...
page 66)
- the ability to CHANGE, IMAGINE, and CREATE
Tu) Let me see, if I take all my known ideas, I think...
- the ability to SEE and INTERPRET signs
Tu) This changes the relationship between an object and it's image
So) Amidst the signs and objects that surround the machine, some are related to the OUTSIDE WORLD, they are NON-SELF and others, SELF, are part of the machine. With the existence of this SCHEME, you can have SELF-AWARENESS.
Tu) Let's see which part is me and which isn't me.
Pe) Without knowing ahead of time
Baby) Translation: Look, where am I and where's NOT I?
page 67)
An) Soon, machines will be able to fix themselves, use themselves, and maybe even EVOLVE themselves.
Sn) Communicate with each other.
Sn) Exchange ideas.
Dr) When one machine understands another one
robo-tripod) I have quite effective programming
Tu) Let's exchange data banks
CYCLOS
MECHANOS
DYNAMOS
SYDEROS
HELIOS
BENTOS
page 68
IN THE MORE OR LESS FAR FUTURE
Ho) Hi Grandma
Ho) I'm one of the HODOS, researcher.
Gr) Hello HODOS, what problems do you have?
Ho) None, I've come to report my findings. I've brought data of what I've seen, heard and learned.
Gr) Come here! I'll plug you in!
Ho) This always gives me a funny feeling
Ho) I didn't have the time to process everything
Gr) That's why you come here, grandson
Gr) Hmm, I need to update your data analysis program. Open your back port.
Ho) What, again? But only last week...
page 69)
Gr) Oh, do it already!
Ho) haha - it tickles
Gr) That feeling's not tickling, bozo
Gr) He's started to develop problems.
Gr) Now it "tickles", what next?
Sign) ATTENTION SCRAPYARD
Sign) COME FOR A FREE TEST OF CIRCUITS. CHANGE YOUR IQ.
Sign) ARE YOU SURE YOU'RE FUNCTIONAL?
page 70)
Sign) REPROGRAM WITHOUT PROBLEMS
Ho) Hey, look, it's a contest.
Freud) I'm faster than everyone. I caught you. Now I'll make you a deal. If you tell me a TRUTH, I'll let you go, if you tell me a LIE, I'll keep you captive.
Ratbot) You'll keep me prisoner!
Freud) Let me think; if I keep her prisoner, then she's told the truth, and I have to let her go.
Freud) But if I free her, she told a lie! The devil take this!
- He burned out!
Ergos) You know, in my opinion, these first generation robots, one beautiful day...
page 71)
Ergos) Look at that, people!
Ho) I wonder what they're going to do?
Ergos) They're worried by MATHOS
So) Up till now, MATHOS has been working fine. He's always explained his procedures to us.
An) But now, he says we wouldn't understand
Pe) That's absurd!
Dr) MATHOS specializes in N-dimensional manifolds
Sn) At first, his goal was to analyze 4-D space (x, y, z, t)
Sn) And partially, "BLACK HOLES"
page 72)
Ergos) Then, he decided that it would be easier to solve if he changed languages.
Ho) Wow, that's a MUTATION in LOGIC.
Sn) Slip of tougue he-he-he
Ergos) Then things got much farther. Based on his new logic, MATHOS studied and described past and future events in terms of simultaneous causality. That worked better.
Ergos) In truth, MATHOS, seeing that, decided to start over and rewrite all of its operations in the language of BISYNCHRONICITY.
Ho) But people?
page 73)
Ergos) They lost their minds.
Ergos) Now the word "spare part" means nothing to them.
Ho) In truth, a lot baffles me these days.
Ergos) But don't let it worry you. Grandma says we're all going to have to reprogram ourselves with BISYNCHRONICITY.
Ergos) Well, those of us who can be reprogrammed.
Ho) And the rest get the scrap heap
Ho) But what about people?
Ergos) That's the problem. We're not sure what to do with them.
Ergos) Mutation of logic is hard for people
page 74)
So) Anselmi! Did you hear that?
An) What?
Ergos) Let's go recharge?
Ho) Yeah, let's do it.
An) Let me try one more time.
- DISPLAY THE EXPLANATION FOR THE ANALYSIS PROCEDURE OF THE EINSTEIN-ROSENBURG PARADOX!
MATHOS) NO TRANSLATION TO A HUMAN LANGUAGE
An) The devil!
An) There's nothing to do for it.
page 75)
So) Has MATHOS spoken to grandma yet?
An) No, I checked and disconnected it just in case.
An) Because I refuse to tell anything to grandma
So) It's been years since anyone has known how to classify her
So) There's only one thing left to do.
An) You're saying we should just print out everything in MATHOS's memory?
So) I don't trust it anymore, come here and help me...
page 76)
So) Let's go!
THE END
BACK PAGE
Science and Technology in Pictures
Box) This type of publication is printed for the first time in USSR. They answer serious questions about science and technology, even though they are in the category of picture books. We present to the Soviet reader the books of J.P. Petit, "What do robots think about?" and Stuart's "Secret Catastrophes." Publisher "Mir", based on the fact that in France, where these books first saw light, they enjoy popularity among scientists, as well as curius youths. The publisher would be grateful to the reader for comments on this form of popularization of new directions in science and technology.
PUBLISHER MIR
Translator into english: Roman Zadov
This work has been generously sponsored by The Burnside Institute for Higher Research
This is a copy of "What do robots think about?" by the famous French author Jean-Pierre Petit. Originally written in 1982, and translated into russian in 1987, this book remains highly relevant in the field of robotics (the first chapter could be the inspiration for the Roomba). Packaged in the form of comics, these concepts were a lot of fun for me as a child, and I hope you will enjoy them too.
This is a literal translation, and some ideas are sure to have been lost, so for another translation, as well as many ther books, and languages, visit the author's site.
FRONT COVER:
WHAT DO ROBOTS THINK ABOUT?
1) Look at that, people!...
2) I wonder what they're going to do?
Inside cover 1
Pelican) We expect that robots don't think...
Inside cover 2
Snail) Robots can do anything, but what do they think about it?
(publishing pages translation omitted)
PROLOGUE (page 7)
Pe) Anselmi, what are you constructing?
Sn) Is it a ship?
An) You'll see soon. This is.. the... hull
Dr. Bird) This is no ship!
Sn) Excellent! It's a steam-powered pelican!
Pe) Hwa-ha?
page 8)
Sn) Yep, that's the thing! It even waddles like a pelican
Pe) Oh, sure, Terezie, when I can be made the butt of the joke, you're ready to oblidge!
- Ah, look at that...
Dr) A hovercraft snail!
< Klok, klok, klok >
Sophie) Oh, great, look at that, more work!
page 9)
An) So what's wrong with that?
So) Once again, you've turned our living room into a laboratory. Everythig's turned upside-down!
Dr) So long ...
An) Stupid device
- all finished
- It's so hot around here
- Hmmmm
Pe) Anselmi's completely broke his vaccum cleaner.
Dr) Yes and no, he's optimizing it
page 10)
PROGRAMMABLE AUTOMATA
An) Now I have a programmable vaccume.
- Look; I can turn it on and control it with the help of these levers...
Dr) So what's the difference? Instead of controlling it with your hands, you now have to control it with these levers?
An) Left! Right !
- When the work is done, I drive it to a device that recharges its battery.
Dr) So, what's so interesting about this?
An) I position my TURTLE by the bin, where she deposits the dust.
page 11)
An) When the work is done, I drive it to a device that recharges its battery.
Dr) So what's so interesting about this?
An) You'll see in a second. I'm going to press this lever,
- and my Turtle can redo the same path, going around the furniture!
page 12)
An) And then the dust empties where it's supposed to!
Dr) And then it returns to recharge ...
And) Let me reveal the mystery...
(Clockwise, starting on the very top left)
- levers for mannual operation
- contacts for recharging
- battery
- punch tape
- turbine of the vaccuum
- freely moving front wheel
- independently controlled back wheels
- electromagnet for opening the dust box
An) When I turn on the Turtle, this tape begins to unwind with a constant speed. it has commands written on it as perforations.
- In this way, the entire task is written in the MEMORY of the Turtle.
page 13)
An) Next, we'll attach the Turtle to a charging station that rewinds the tape
Dr) This means that you can write a program and make the Turtle understand and execute commands? But where did you find the tape?
An) I took it out of an old piano
- Cleanup in the kitchen!
- ?
- Sophie, watch out!
page 14)
An) Now we can go out, and the turtle can take care of cleaning the house.
So) Good job Lanturlu
(Three hours later..)
So: My livingroom! What a disaster!
An) Oh-oh. I'm about to get it...
(Analysis of the situation allows us to deduce what happened.
- the Turtle encountered the cat's tail, but this was not forseen in its programming.
- She immediately vaccumed up the tail.
Cat) Meeeeoooow!
page 15)
- In a terrible panic, the cat dragged the robo-turtle on its tail around the room - destroying all in his wake.
Rat) Help!
So) The acid from the battery spilled on the rug and ate through it. The Turtle even sucked up my tights.
An) Ok, ok, I'll start cleaning it up.
page 16)
SYSTEMS WITH INPUTS AND OUTPUTS
So) The cat left a note, "Farewell, I cannot remain another moinute in this madhouse. Do not expect me to catch mice for you anymore."
Su) Without a warning!
An) My Turtle, that I thought was pretty clever, turns out to be very dumb.
So) Why would it be smart? it doesn't know what's going on around it! It's imperative that we supply it wish sensors.
An) You are right, we need to make my turtle see. This device is sensitive to INFRARED, i.e. heat radiation.
Sn) And in what way will it react to heat?
- heat radiation
page 17)
An) With a couple of hundred of these elements, I have created a basic "eye," similar to the infrared eyes of a snake.
An) Our tile floor is cold. For the Turtle, this will be baseline temperature. In this way, it will be percieved as black.
An) Our walls, which have paper wallpaper, are slightly warmer, and so they'll be percieved as gray.
Sn) The field of vision is divided into cells to ease perception.
Page 18)
- The field of vision of the Turtle is a cone.
Dr) The Turtle can tell the angle between an object and its optic axis.
Dr) Each infrared element corresponds to a pixel MxM sized 1(degree)x1(degree), which determines the resolution power of the optical system; the image is RASTERIZED that is, turned into white, black, and gray rectangles.
An) I've junked all the old programs, and I plan to give the Turtle a simple reflex that will allow it to avoid obstacles. If one of the photosensitive elements in the turtle's eye recieves radiation whose intensity exeeds the value of the floor; and if this source of heat is below the ocular axis, and the angle phi is less than phi(min), the Turtle will take a 90 degree turn to the right.
Box) When the intensity of the heat radiation is above the threshold, and angle psi is less than dthe horizon value of psi(min),
Dr) That means the signal is in the shaded part of the circle
Tr) So I turn
page 19)
So) In this manner, the Turtle represents a SYSTEM WITH INPUTS AND OUTPUTS
-"BLACK BOX"
Sn) What's in this "black box"?
Dr) A PROGRAM that analyzes the input signal (Information processing net) and on that basis, determines the future actions, for example turning and moving right.
An) Now, my turtle will suck up dirt and won't run into anything. She will avoid walls, and also furniture - which is even darker than the floor.
An) It's all in order. The problem is solved. Before, it was all off, but now I'm sure I've got it straight.
Pe) Anselmi!!!
An) What?
Pe) The Turtle has sucked up Terezin!
An) Seriously? Byt why?
page 20)
Dr) Terezin is cold blooded and also her body is positioned ner the floor. That is why she had the same darkness as the floor and was invisible to the turtle.
Sn) Careful!
An) I connected a candle to her back. Now she'll be safe.
So) Look, when the wall is at an acute angle, as here, the turtle has to turn twice. That's kind of awkward.
An) Yeah, it would be better to turn left here.
SIGNAL ANALYSIS
An) How should I do this?
- I know!
Box) OLD PROGRAM
If a warm object appears in region A, turn right.
Box) NEW PROGRAM
If a warm object appears in region A, turn right.
If it appears in region B, turn left.
page 21)
Sn) Ok, but what if the turtle comes to the wall at a right angle?
Pe) Anselmi, she'll burn up!
An) Dumb animal!
Dr) The solution is this: enter into the turning algorithm - if the signal demands a simultaneous turn to the right and left, turn 90 degrees right.
An) Sophie! I've thought of how to improve the interpretation of the signals recieved by the eyes of the Turtle. For example, if this object comes into its field of vision, she only has to count up the number of elements comrising the object in field A(left) and in field B(right).
So) You also have to forsee the Turtle being able to go between objects.
page 22)
An) For that, It is sufficient to reduce zones A and B in the display of the turtle.
REACTION TIME
An) A device is interesting if it works fast. I'm going to maximize the speed of the turtle so that it does the work fastest.
- Oh - no! Not again!
page 23)
So) Anselmi - you can't expect the system to react to inputs immediately. Every self-controlled system has REACTION TIME.
- Because of reaction time, you can't catch the bank-note when I let it go.
An) I missed it again!
Rat) It's imperative to be devilishly fast.
Box) Attention! This Input/Output system has a short response time.
An) Cleanup takes a lot of energy. My Turtle is completely spent.
Sn) The turtle should automatically return to recharge.
Dr) Sure, but how?
page 24)
FEEDBACK SYSTEMS
An) When the charge in the battery decreases, so does the voltage. It's easy to include, in the program, something like this: IF the voltage is below a certain level, THEN return to recharge. There's still a question - how does the turtle find the charges?
- Between projects, I made a device that can recharge itself using solar rays.
Sn) We can call it a POWER PLANT
An) This wire can be heated to direct the turtle, but how?
So) You have to include FEEDBACK into the Turtle.
page 25)
So) The wire should be be seen as an obstacle. We should hang it up as high as possible.
Pe) And so, the potential will show up in the top segment of the display.
Sn) It's all so simple! If the charge of the batteries falls below a certain voltage, th eTurtle will start looking for a heat source. There are two possibilities: Either the coil is in the field of vision and she needs to drive to it, or it isn't and the turtle should turn until it is.
So) As soon as the coil is visible, the turtle will drive to it with the following program:
Box) Correlate the direction of movement proportionally to the size of the angle and in the direction of the angle. This is called LINEAR FEEDBACK.
page 26)
- With a change of course, the angle psi also changes. This is how ANGULAR direction works.
Tu) Turn just - but to the right.
Dr) This is how the turtle directs itself to the target.
ORDER OF FEEDBACK SYSTEMS
So) Hey Anselmi: Let me present you with a game. On my signal, you have to put the point of your pencil on the point of mine.
An) Aha, you want to say that I present a feedback system - "black box"
page 27)
So) Some sort of "TERMINATOR"
An) I direct my pencil with the observable angular difference.
So) Come on, faster!
An) What's going on here? I can position my pencil but only after a couple of correlations (attempts).
So: This happens because of INERTIA
Dr) In a FIRST ORDER SYSTEM, the direction command works directly on speed without regard for inertia. In this situation, there are no vibrations.
In a SECOND ORDER SYSTEM the direction commands (expressed as a force) act on acceleration. Thanks to inertia, this can lead to oscillations.
page 28)
Dr) A person, thus, is a 2nd order system. The flush tank in a toilet is a good example of a first order system. Speed of the water rising is proportional to the height (e) needed to rise.
Pe) In truth, this is so, the height of the water tank does not vibrate.
Dr) But due to the physical nature of things, INERTIA appears all the time. If they're pushed a little, a first order system can turn into a second order one.
page 29)
Sn) Sophie, Leon said that I move so slow that I'll never feel inertia, and so I must be a first order system.
SELF-REGULATING SYSTEMS
Dr) Let's return to our turtle
Dr) Anselmi made a clever system. The Power Plant creates a difference in potential between two circular wires. The copper wheels of the turtle bridge the contact. As soon as the front wheel hits the anode, and the back, the cathode, the turtle stops and charges. When charged, turns around and returns to work. When it runs out again, the heated coil directs the turtle to the charging station.
page 30)
So) You also stop eating when the pressure on the walls of your stomach exceede a certain point.
An) Who me..?
Sn) Yeah, you. The stomach works exactly the same way as the toilet bucket.
Dr) No, the toilet bucket acts like a bucket.
Pe) Ugh, stop being so vulgar.
So) These are examples of systems that seek to stay within their parameters by regulating between a minimum and maximum value.
An) I propose that when I eat, I seek to regulate the amount of sugar, salt, etc... in a balance between some maximum and minimum value.
page 31)
An) So then why should I make something similar in a machine?
So) I'd say the opposite - this machine is similar to a person.
- Everything that's made by people; techniques and technology - are ways of recreating or mimiking natural processes.
- After humans learned how to copy the forms of nature, they started copying the methods.
Box) An Input/Output system with a quick response
page 32)
So) A living organism - it's a fantastic machine; it's self-regulating and homeostatic; it maintains the right amounts of water salt, blood, muscles, and also its own form.
- Cells also stop their growth when they come into contact
Dr) Any noticeable deviation of values from the standard results in the destruction of tissues.
Sn) Any disruption of the cellular cover results in increased cell division on the sides of the wound.
Dr) The division of cells stops as soon as contact between them is restored.
page 33)
Dr) If this process of healing stops acting too slowly, inflamation will occur.
Pe) What are you doing?
An) Sophie asked me to rid the house of mice. I have to catch them all.
I'll make the turtle an arm that coomerates with its eye.
Dr) Mice are warm. 42 degrees c
An) This is necessary for determining the baseline. The floor is "black", the walls are "grey", the mice are "white". We just have to edit the program to chase any object that's hot enough (higher than the constant C which is written on the magnetic tape)
page 34)
An)First it fids a mouse
M) What kind of thing is this?
An) Then the turtle aims from above.
Dr) Then it nears until the mouse is in the zone of capture.
M) Let me go! What is this!
An) There we go.
page 35)
So) Anselmi!
An) What?
Sn) Is this about me?
So) Can you make this thing let go of the cup, or at least what's left of it?
An) Of course, it can't tell a mouse from a cup of warm coffee
Snake) The devil!
Dr) The same thing can happen to a snake when she goes hunting at night
An) My Turtle can get confused by a nearby target and a far one. The intensity of the radiation varies with the inverse square of the distance to it.
Sn) The turtle is nearsighted!
page 36)
So) Don't forget that when winter comes, your Turtle will be totally blind.
An) Why?
So) Because we have a heated floor.
An) The situation is hopeless.
Dr) Maybe you can make it hybernate
Sn) We need to teach the Turtle to RECOGNIZE SHAPES of objects to tell them from one another.
Pe) While you're at it, why not teach it to read?
page 37)
PATTERN RECOGNITION
An) This video camera with a high resolution will fit us a lot better than the infrared eye.
So) The display of any object that's in front of the camera consists of a collection of small squares with x and y coordinates.
Dr) For some shape to be recognized, it needs to be known
Dr) Let's start by teaching the Turtle to MEMORIZE SHAPES
page 38)
Dr) For example, letters of the alphabet
Pe) It's clear to me what needs to be done. The letters need to be shown to the machine one after another, and it needs to compare them to images that it knows.
Dr) The signal being analyzed should be overlayed a known signal - one that is stored as a negative.
An) That way, if the signals correspond, the field will be consistently grey.
Page 39)
Dr) Actually, each image is represented by a collection of ones and zeroes that are organized by their x and y locations. The integrated computer will save all the differences and coincidences. Coincidences - 4 cases. Number of cells in signal - 7. Ratio of coincidences to cells - 4/7
Pe) Sure, but there's still a lot of problems. The turtle can't recognize a letter if it's tilted or at the wrong distance.
Pe) Are you relying on good luck?
An) The devil! What should I do?
So) I know!
page 40)
So) I find the center of gravity of each symbol
So) Before recording the symbol in the memory of the machine, I'm going to have to record its center as well.
So) That way, whenever a symbol enters the field of vision
S0) I can immediately make it compute the center of the part that it sees, G1, and then align the center of the camera with the center of the image.
page 41)
So) When this is done, the machine can compute the center of G2 the new image that it now sees, and then allign with it.
So) In this way, it can hone in on the center of the object.
Pe) But we're still left with the problem of the size of the object.
- The integrated computer can make N copies of the image, increased or reduced.
So) And then compare these images to the saved ones in the memory.
Tu) Oh yeah, it's a P!
page 42)
An) Wait a second! We can get rid of these systematic comparisons of images. From afar, each letter looks like an indistinct object. Each of these images has a center and a CHARACTERISTIC DIAMETER
Sn) But how are you going to determine this diameter?
An) I'm going to take each of the darkened squares M and find the distance from them to the center G. Next, I'm going to add all the lengths GM and divide by their number. This average of lengths gives me a radius R and then D becomes 2R.
An) That way we can assign a characteristic diameter to each letter and symbol.
An) Instead of memorizing all the symbols, I'm going to give them all a center of Xg = 0 and Yg=0 - and the same Diameter, for example D = 1.
Sn) In other words, you center and focus the shape.
page 43)
So) In this way, we've established, that the Turtle can automatically find the center of the image, then by averaging points find it's diameter D. If D is different from 1, the Turtle's computer can shrink or grow the image until the Diameter is 1.
Sn) Now all that's left is to go through all the images in memory and look for a match.
page 44)
Sn) But what if we include the possibility that the image can be tilted with respect to the horizon? Then we'll have to turn each image from 0 to 360 degrees.
Pe) If people worked that way, each hard job would have to be done at each angle as well! This would take much more time.
So) If we try to do this with only one MICROPROCESSOR, like the Turtle, then you're right. But you can't say the same about a system of thousands of microprocessors, all working IN PARALLEL.
-...No, that's not it
-...No, not it
- Hey guys! I found it!
Sn) The time for INFORMATION PROCESSING is a lot smaller.
page 45)
Dr) A person can recognize letters at reading speed. Your brain can decode information that's been sent by your eyes as fast as 10,000 microprocessors working in parallel.
Sn) And really, it's a wonder when a person memorizes an image, he's ONE.
Dr) Boggles the mind
So) General pattern recognition is a much more difficult problem. We've only shown a bare outline of the problem.
Pe) Is this the first Turtle that can read?
Pe) Would you look at that...
page 46)
Dr) If the turtle can recognize letters, then it can recognize groups of letters, words, and whole phrases.
Box) DANGER
An) We're about to see a very interesting experiment.
Sn) It's snowing
Box) This is the image of a letter that the Turtle can see at a certain moment.
page 47)
NOISE
Dr) Each letter is disturbed by BACKGROUND NOISE. First idea: There's no 100% correlation to any letter.
An) Despite the noise, I can still tell these letters apart. Even though they're no more than 75% complete.
Sn) But here, we can't tell anything.
So) But we can superimpose many images of the object on the same location.
Dr) Since you have two eyes, at any moment, you're looking at a superimposition of two images.
page 48)
IMAGE PROCESSING
So) A combination of N images can improve the signal, making it more clear.
Dr) We can, for example, put them on top of each other and average.
So) There are many more ways to combine images with mathematical methods.
An) When I block the signal with my fingers, I can't read it.
An) But when I move my hand back and forth, I can!
page 49)
Pe) If I understood you correctly, you want to teach your Turtle to recognize shapes and read with noisy conditions. But object recognition is dependent on finding it's center and diameter. The BACKGROUND NOISE makes that problematic...
So) I have an idea how to "clean up" the image. We can examine each pixel and and eliminate each one that doesn't have at least two neigbors.
So) This will let is get rid of a large number of isolated points of noise.
Dr) Then you can invert the image and do it again.
Sn) Here's the result of the second "clean up"
page 50)
Pe) Fine, so let's say the Turtle can read in any weather? Why do we need that?
Dr) Imagine, Leon, that the internal systems of the camera become similar to a human's.
Sn) The number of pixels at the center of the view will become much greater than on the edges.
WHEN READING TEXT the axis of sight moves
around the symbols erratically only a few parts of the image are shown clearly,
but the mind reconstructs enough of the text to find a MOST LIKELY sentence.
So) This lets us read very rapidly.
page 51)
So) In truth, GATHERING all the information, going from word to word - letter after letter - element after element - is unending work
Dr) To tell symbols apart, it's enough to see "key" forms of the image
Pe) I see. That information is extra
An) In most cases, it's enough to look at the beginning, but not the end of the word.
An) And what you can say about reading, you can say about any visual perception.
An) When a person thinks that there's something wrong with his image, his eye looks over the same area many times quickly.
page 52)
PERCEPTION
Dr) Our sensory organs are always getting information about the surrounding world.
So) But from all the information we recieve, we only process a MINIMUM required to tell apart shapes, sounds, etc.
So) These images are then compared to the enormous bank of images and sounds we have in our memory...
- memory written in the sand
Sn) Which are constantly updated
Dr) It's determined that if you deprive somebody of all information from the outside world, he will soon become UNNORMAL.
page 53)
Sn) After a few days of total sensory deprivation, he can't complete even basic motor movements, like for example, holding a glass. This gives the impression that our impressions of the outside world need constant reinforcement.
Dr) We are in a state of constant learning.
Pe) Alright, but let's go back to our Turtle. She can quickly and constantly recognize different shapes. But WHAT NEXT?
Pe) She's still dumb.
page 54)
Pe) She can only do what's written in her memory.
Signs) Enter, Stop, Exit, Food, Attention
So) We can add a generator of random signals to make the actions a little different.
Sn) But it will at some point have to make desicions.
Dr) All living things have systems for making desicions.
Pe) I have to admit, I sometimes have to do things... that I don't understand...
page 55)
INTELLIGENCE AND STUPIDITY
Pe) This is all illusion. Every machine is and will always be completely stupid.
so) Leon, what is intelligence?
Pe) Intelligence... I have to quickly find a definition! I have to admit I'm not a person...
An) Often, we act like a real mechanism, as if programmed...
Sn) A person that can only march to a beat doesn't need a brain. A spinal chord is quite sufficient. *
* Albert Einstein said this
page 56)
An) Intelligence - this is a thing that floats on top of an ocean of reflexes and goals.
So) What we've done so far is try to mimic the behavior of animals on different levels of evolution.
So) But even though no newborn can speak or dance, still from the moment of birth, he has a genetically PREPROGRAMMED INSTRUCTIONS of behavior, i.e. INSTINCTS.
Dr) The teach, in part, to have an ambition to explore and to learn.
An) I'd like to teach my turtle that way. So that it would be ameable to learning.
page 57
METHOD OF TRIAL AND ERROR
Tu) Interesting, this is a word that's unknown to me. But it looks like the arrow is showing a direction.
Tu) I'll enter this word into memory
Box) STAIRS
Tu) Let's check it out...
page 58)
Tu) Now I know that I don't like stairs.
- The turtle can carry out experiments and then use the results to its own benefit.
Tu) At sixty volts I can charge very fast, but I overheat.
So) It would be nice to know what we can learn with this method of learning.
page 59)
Dr) A machine can store any type of data in its memory.
Reels) Chess rules / Chess games 1890-1912 / Chess games 1913 - 1922
Sn) First, you need access to a DATA BANK
So) Then, after using its sense organs
Dr) These data can be ANALYZED and connections can be found between them.
Tu) I've noticed that the chance of rain increases when dark masses gather in front of the sun.
page 60)
Tu) And rain is bad for my circuits.
Sn) These continual experiments cause the Turtle to keep changing its STRATEGY.
Sn) Ah, what a wonderful weather we have today.
Tu) Hmmm, let me think, last time I was in this situation, a rook move was a bad idea.
Tu) Using this new information, and the game between Alekhine and Morphy in 1924, the pawn looks like it can determine the game.
Tu) Let's try that operation
Tu) Let's see...
page 61)
So) O - ho - ho
Sn) And so, where does intelect begin and stupidity end?
TURING TEST
- a "mathematician" developed a test for intellect
Sn) We shouldn't forget that after 1981 the chess champion has become the computer.
An) Hello? Mademuselle, you say that there's a spot on the 22:30 train?
Telephone) Yes Monsur, I'll reserve you a seat.
An) Did you know you have a charming voice?
Telephone) Please Monsur, be serious.
An) Are you free some night this week?
Telephone) Monsur, this is impossible
An) But why is that?
page 62)
An) Hmm, let me try a test.
An) What is your opinion on waffles?
Telephone) Waffles? What county is that in?
An) Again, it's one of those damn machines
Telephone) Hello! Monsur? Your reservation!
Box) A machine can be considered intelligent if its behavior cannot be distinguished from a person's. - Turing
Pe) Thank god, there's still a long way to go.
Pelican 2) Artificial intelligence? Ptui!
page 63)
Pelican 2) All of this is just tricks. Nobody is going to convice me that one day, a computer can compare intelligences with...
Sn) A bird-brain?
Pe) O-o-o watch it!
Sn) he-he-he
Sn) We are in danger of not noticing when one beautiful day, machines get intelligence, and we will not be ready for it.
Hand) Terezia!
Dr) What bunk!
page 64)
ARTIFICIAL INTELLIGENCE
An) How to get out?
An) You just mentioned, Sophie, that we've started making machines that simulate our living things.
Pe) That's not to be taken literally. Even a car resembles a horse.
So) Yes, but both a car and a horse take in chemical energy, and convert it into mechanical energy of movement, and so the are FUNCTIONALLY engines.
Sn) Yes, that's true, a machine copies and complements nature.
page 65)
-With the help of CYBERNATICS and INFORMATION SCIENCE, humans have imitated the functions of REGULATION and PROGRAMMING living things, imitating instincts.
So) For scrap metal!
Tu) Run for it!
Sn) Instinctive response
Tu) It's getting cold. I'll turn on the heating.
Pe) It's hard to stop on this good path
Dr) People have also been able to give machines:
Dr) Ability to LEARN
Sign) Bridge out. Detour 500m
- Ability to ASSOCIATE
Tu) If all greeks are liars, and if Epimenides is greek...
page 66)
- the ability to CHANGE, IMAGINE, and CREATE
Tu) Let me see, if I take all my known ideas, I think...
- the ability to SEE and INTERPRET signs
Tu) This changes the relationship between an object and it's image
So) Amidst the signs and objects that surround the machine, some are related to the OUTSIDE WORLD, they are NON-SELF and others, SELF, are part of the machine. With the existence of this SCHEME, you can have SELF-AWARENESS.
Tu) Let's see which part is me and which isn't me.
Pe) Without knowing ahead of time
Baby) Translation: Look, where am I and where's NOT I?
page 67)
An) Soon, machines will be able to fix themselves, use themselves, and maybe even EVOLVE themselves.
Sn) Communicate with each other.
Sn) Exchange ideas.
Dr) When one machine understands another one
robo-tripod) I have quite effective programming
Tu) Let's exchange data banks
CYCLOS
MECHANOS
DYNAMOS
SYDEROS
HELIOS
BENTOS
page 68
IN THE MORE OR LESS FAR FUTURE
Ho) Hi Grandma
Ho) I'm one of the HODOS, researcher.
Gr) Hello HODOS, what problems do you have?
Ho) None, I've come to report my findings. I've brought data of what I've seen, heard and learned.
Gr) Come here! I'll plug you in!
Ho) This always gives me a funny feeling
Ho) I didn't have the time to process everything
Gr) That's why you come here, grandson
Gr) Hmm, I need to update your data analysis program. Open your back port.
Ho) What, again? But only last week...
page 69)
Gr) Oh, do it already!
Ho) haha - it tickles
Gr) That feeling's not tickling, bozo
Gr) He's started to develop problems.
Gr) Now it "tickles", what next?
Sign) ATTENTION SCRAPYARD
Sign) COME FOR A FREE TEST OF CIRCUITS. CHANGE YOUR IQ.
Sign) ARE YOU SURE YOU'RE FUNCTIONAL?
page 70)
Sign) REPROGRAM WITHOUT PROBLEMS
Ho) Hey, look, it's a contest.
Freud) I'm faster than everyone. I caught you. Now I'll make you a deal. If you tell me a TRUTH, I'll let you go, if you tell me a LIE, I'll keep you captive.
Ratbot) You'll keep me prisoner!
Freud) Let me think; if I keep her prisoner, then she's told the truth, and I have to let her go.
Freud) But if I free her, she told a lie! The devil take this!
- He burned out!
Ergos) You know, in my opinion, these first generation robots, one beautiful day...
page 71)
Ergos) Look at that, people!
Ho) I wonder what they're going to do?
Ergos) They're worried by MATHOS
So) Up till now, MATHOS has been working fine. He's always explained his procedures to us.
An) But now, he says we wouldn't understand
Pe) That's absurd!
Dr) MATHOS specializes in N-dimensional manifolds
Sn) At first, his goal was to analyze 4-D space (x, y, z, t)
Sn) And partially, "BLACK HOLES"
page 72)
Ergos) Then, he decided that it would be easier to solve if he changed languages.
Ho) Wow, that's a MUTATION in LOGIC.
Sn) Slip of tougue he-he-he
Ergos) Then things got much farther. Based on his new logic, MATHOS studied and described past and future events in terms of simultaneous causality. That worked better.
Ergos) In truth, MATHOS, seeing that, decided to start over and rewrite all of its operations in the language of BISYNCHRONICITY.
Ho) But people?
page 73)
Ergos) They lost their minds.
Ergos) Now the word "spare part" means nothing to them.
Ho) In truth, a lot baffles me these days.
Ergos) But don't let it worry you. Grandma says we're all going to have to reprogram ourselves with BISYNCHRONICITY.
Ergos) Well, those of us who can be reprogrammed.
Ho) And the rest get the scrap heap
Ho) But what about people?
Ergos) That's the problem. We're not sure what to do with them.
Ergos) Mutation of logic is hard for people
page 74)
So) Anselmi! Did you hear that?
An) What?
Ergos) Let's go recharge?
Ho) Yeah, let's do it.
An) Let me try one more time.
- DISPLAY THE EXPLANATION FOR THE ANALYSIS PROCEDURE OF THE EINSTEIN-ROSENBURG PARADOX!
MATHOS) NO TRANSLATION TO A HUMAN LANGUAGE
An) The devil!
An) There's nothing to do for it.
page 75)
So) Has MATHOS spoken to grandma yet?
An) No, I checked and disconnected it just in case.
An) Because I refuse to tell anything to grandma
So) It's been years since anyone has known how to classify her
So) There's only one thing left to do.
An) You're saying we should just print out everything in MATHOS's memory?
So) I don't trust it anymore, come here and help me...
page 76)
So) Let's go!
THE END
BACK PAGE
Science and Technology in Pictures
Box) This type of publication is printed for the first time in USSR. They answer serious questions about science and technology, even though they are in the category of picture books. We present to the Soviet reader the books of J.P. Petit, "What do robots think about?" and Stuart's "Secret Catastrophes." Publisher "Mir", based on the fact that in France, where these books first saw light, they enjoy popularity among scientists, as well as curius youths. The publisher would be grateful to the reader for comments on this form of popularization of new directions in science and technology.
PUBLISHER MIR
Translator into english: Roman Zadov
This work has been generously sponsored by The Burnside Institute for Higher Research