Marvin Minsky "The Emotion Machine": Chapter 8.1-2 "Creativity"
8.1 Creativity
βAlthough such a machine could do many things just as well and perhaps better than we do, in another it would certainly fail, and it would be found that it does not act consciously, but only due to the location of its organs.β β Descartes. Reasoning about the method. 1637
We are accustomed to using machines that are stronger and faster than humans. But before the advent of the first computers, no one guessed that a machine could do anything more than a limited number of different actions. This is probably why Descartes insisted that no machine could be as inventive as man.
βFor while the mind is a universal instrument, capable of serving under the most varied circumstances, the organs of the machine require a special disposition for each individual action. Hence, it is inconceivable that there should be so many different arrangements in the machine so that it could act in all cases of life in the way that our reason makes us act. β Descartes. Reasoning about the method. 1637
In the same way, it was previously believed that there was an unbridgeable gulf between man and animals. In The Descent of Man, Darwin remarks: "Many authors have insisted that man is separated by an insurmountable barrier from the lower animals in regard to mental faculties". But then he clarifies that this is the difference "quantitative, not qualitative".
Charles Darwin:βIt seems to me now fully proven that man and higher animals, especially primates ... have the same feelings, impulses and sensations; all have the same passions, affections and emotions, even the most complex ones, such as jealousy, suspicion, competition, gratitude and generosity; ... have, although to varying degrees, the ability to imitate, pay attention, reason and choose; possess memory, imagination, association of representations and reason.
Darwin further notes that "individuals of the same species represent all stages, from sheer stupidity to great intelligence" and claims that even the highest forms of human thought could develop from such variations - because he sees no insurmountable obstacles to this.
βOne cannot deny at least the possibility of this development, because we see daily examples of the development of these abilities in every child and could trace completely gradual transitions from the mind of a complete idiot ... to the mind of Newtonβ.
Many people still find it difficult to imagine the transitional steps from the animal to the human mind. In the past, this view was forgivable - few thought that just a few small structural changes can greatly increase the capabilities of machines. However, in 1936, the mathematician Alan Turing showed how to build a "universal" machine that could read the instructions of other machines and then, by switching between those instructions, could do everything that those machines could do.
All modern computers use this technique, so today we can use one device to organize a meeting, edit texts or send messages to friends. Moreover, once we save these instructions inside machines, programs can change so that the machine can expand its own capabilities. This proves that the limitations that Descartes observed were not inherent in machines, but were the result of our old-fashioned ways of building or programming them. For every machine we have designed in the past, there has been only one way to accomplish each particular task, while a person, if he has difficulty in solving a task, has alternative options.
However, many thinkers still argue that machines will never be able to achieve such heights as writing great theories or symphonies. Instead, they prefer to attribute these skills to inexplicable "talents" or "gifts." However, these abilities will become less mysterious once we see that our resourcefulness could arise from different ways of thinking. Indeed, each previous chapter of this book has shown how our mind offers these alternatives:
Β§1. We are born with many alternatives.
Β§2. We learn from Imprimers (Imprinters) and friends.
Β§3. We also learn what not to do.
Β§4. We are capable of reflection.
Β§5. We can predict the consequences of imagined actions.
Β§6. We use vast stocks of common sense knowledge.
Β§7. We can switch between different ways of thinking.
This chapter discusses the additional features that make the human mind so versatile.
Β§8-2. We look at things from different points of view.
Β§8-3. We have ways to quickly switch between them.
Β§8-4. We are able to learn quickly.
Β§8-5. We can effectively recognize relevant knowledge.
Β§8-6. We have different ways of representing things.
At the beginning of this book, we noted that it is difficult to perceive oneself as a machine, since no existing machine understands the meaning, but only executes the simplest commands. Some philosophers argue that this is how it should be, because machines are material, while meaning exists in the world of ideas, a region outside the physical world. But in the first chapter, we suggested that we ourselves limit machines by defining meanings so narrowly that we cannot express their diversity:
βIf you 'understand' something in only one way, you are unlikely to understand it at all - because when something goes wrong, you hit a wall. But if you imagine something in different ways, then there is always a way out. You can look at things from different angles until you find your solution!β
The following examples show how this diversity makes the human mind so flexible. And we will start by estimating the distance to objects.
8.2 Distance estimation
Do you want a microscope instead of an eye?
But you're not a mosquito or a microbe.
Why watch us, judge for yourself,
On the aphids, neglecting the heavens β A. Pope. Experience about a person. (translated by V. Mikushevich)
When you are thirsty, you look for something to drink, and if you see a mug nearby, you can just take it, but if the mug is far enough away, you will have to go to it. But how do you know what things you can reach? A naive person sees no problem here: "You just look at the thing and see where it is". But when Joan spotted an approaching car in chapter 4-2 or grabbed a book in 6-1, how did she know the distance to them?
In primitive times, people needed to evaluate how close a predator was. Today, we only need to evaluate if there is enough time to cross the street - nevertheless, our life depends on it. Fortunately, we have many ways to estimate the distance to objects.
For example, an ordinary cup is the size of a hand. So what if a cup fills as much space as your outstretched hand!then you can reach out and take it. You can also estimate how far a chair is from you, as you know its approximate size.
Even if you don't know the size of an object, you can still estimate its distance. For example, if one of two things of the same size looks smaller, then it is further away. Such an assumption may be erroneous if this thing is a model or a toy. If objects overlap, regardless of their relative sizes, the one in front is closer.
You can also get spatial information about how parts of a surface are lit or shaded, as well as the perspective and surroundings of an object. Again, such clues are sometimes misleading; the images of the two blocks below are identical, but the context suggests that they are of different sizes.
If you assume that two objects lie on the same surface, then the one that lies above is further away. Finer-grained textures appear farther away, as do blurry objects.
You can estimate the distance to an object by comparing different images from each eye. By the angle between these images, or by small "stereoscopic" differences between them.
The closer an object is to you, the faster it moves. You can also estimate the size by how quickly the focus of vision changes.
And finally, besides all these perceptions, you can estimate the distance without using your vision at all - if you have seen an object before, you remember its location.
Student: why so many methods, if two or three are enough?
We make hundreds of distance estimates every waking minute, and yet we almost fall down stairs or crash into doors. Each method for estimating distance has its drawbacks. Focusing only works on close objects - some people can't focus at all. Binocular vision works over long distances, but some cannot match images from each eye. The other methods don't work if the horizon isn't visible, or texture and blur aren't available. Knowledge applies only to familiar objects, but an object can be of an unusual sizeβnevertheless, we rarely make fatal mistakes, since we have many ways to estimate distance.
If each method has its pros and cons, which one to trust? In the following chapters, we will discuss several ideas about how we are able to switch between different ways of thinking so quickly.
Thanks for the translation katifa sh. If you want to join and help with translations (write in a personal or email [email protected])