Artificial intelligence ( AI ) and machine learning are deeply inspired by how humans learn and interact with their environment or society. Reinforcement learning supervised learning and unsupervised learning have all been discovered by studying how the human brain actually LEARNS.
I will try to bring about a basic of how about we would develop the AI society which will be based on how we, humans, developed. Here, I will refer to the AI society as computers. I think the whole development to be broken down to 3 ideas
Survive, Develop, Reproduce…
If both learners, humans and computers, want to develop their intelligence, why are doing it? What’s the reason which attracts them towards development?
We, humans, have a kind of curiosity to explore the concepts and ideas which are unknown to us. This is because our brain is evolved in such a manner that it needs to explore, discover or invent new ideas as it is the pathway to survival. We develop so that we can survive.
Well, computers still don’t have the ability to find new ideas or concepts and they do exhibit curiosity. We need to make them explore new things ( or data ) so that they could generalise themselves in some particular tasks. Once, they will reach the level where they could make actions in favour of their survival, then they would achieve curiosity.
Computers, since driven by mathematical frameworks, could have extremely high development, be it emotional, physical or scientific. Mathematics, if applied effectively, could easily surpass human intelligence even though we created it. See the AlphaGo AI to believe.
How will a Computer initially start to learn and develop itself? Will it be similar to a human?
We, right from our birth, interactive with other humans, and thereby learn from them. Parents play a major role here. About 70% of the learning takes place after we take birth. But, every sort of learning, is based on some kinds of rules. These rules could be considered as the base, on which the intelligence thrives. Survival, reproduction, curiosity could be the examples of such rules.
Computers, on the other hand, also need to be given some rules on which they formulate themselves so that they could efficiently work around. A Skin Disease Classifier could never classify a dog or a cat. This is because the classifier is trained or interacted with data which only includes images of skin diseases.
A skin disease classifier could be considered as a human which has never seen anything except for images on skin diseases.
Humans study their origin. Will computers do the same?
Study of origin started as humanity developed slowly and steadily. The study of our birth and origin helps us to understand what mistakes our ancestors made so that we can improve ourselves and hence sustain ourselves.
The AI society may also study their origin ( which they would probably find in mathematics ). If they will be intelligent enough in their survival skills then they would definitely study their origin.
Will AI suffer and have pains just as humans have?
The human brain is a bit dumb in this case. Be it a small pin pricking you, your brain gets the impulses of pain which it thinks would endanger your survival. Every pain and suffering is connected with our survival.
We need to give the AI society some sought of stimuli which could give them a feeling of pain. But, why is pain necessary? This is because pain leads to demands and requirements of remedies. These demands will nurture inventions and discoveries.
Need is the mother of inventions. One who has never been through pains, has never tried something new and will never develop itself.
But, what if an AI does not feel any pain?
No pain simply means no gain. A painless life or functioning cycle will diminish the desire to be alive and survive and develop.
Will, there be crimes in an AI society?
Crimes are caused by the differences between the thinking of two people or a group of people.
Gradual development in the mind of an AI would slowly and steadily widen its thinking capabilities. These capabilities will tend to work in different directions and hence giving rise to different thought processes. The difference in thought processes will ultimately lead to crimes.
Rival thoughts will try to destroy each other as their mind will see the rival as a threat to its existence or survival.
Do we need to programme our AI in such a way that it can reproduce itself?
Reproduction is natural in humans. We have been a kind of programmed to feel attracted towards other humans. In fact, every living organism tends to reproduce. But, why AI needs it?
An AI society will seem more like a society if it expands itself. The mode of reproduction may vary but should be a prime reason for the expansion of society. New ideas and concepts will come up and we will see the same progress as humans did.
A bit of Computer Science?
At the simplest level, survival is the key factor which shaped human society. We need nutrition for survival. For a computer, the turning on of eletricity could be considered as its “birth”. Whereas the turning off of electricity could be termed as its “death”.
If we take this same principle and programme a machine to sustain its nutrition ( electricity in this case ), we can make it more intelligent enough so that it can develop itself.
Imagine a robot sitting and connecting the charging wire, on its own, to its body, when its battery goes low? Well, this has been achieved. In present days, the robots are preprogrammed to charge their batteries when it reaches a certain level. That’s not intelligence.
The above principles could be somehow moulded using mathematics and thereby simulated by a computer. These were purely my ideas and I will like to read our views too. Thank You.