Actual Intelligence | Chapter 1,2,3 & 4

Chapter 1 : Life as we know it.

Our solar system has 7.8 billion human beings, all of them concentrated on a single planet, Earth. Earth offers a perfect environment for life to persist. It feels almost as if someone precisely designed the living environment on Earth.

Since the beginning of time, mankind has been evolving and adapting from one century to another. The civilization we are in right now is a result of centuries of evolution.

Chapter 2 : A little about machines

The most advanced technology known to mankind known at this point of time is AI.

AI is the base of many industries and organizations. Major social media companies like Instagram, Facebook, Pinterest and Twitter etc. Use AI as a tool to draw more users to using their social media platforms. Their core benefit is directly proportional to the amount of time you spend looking at the screen, using their app.

AI makes it possible to get you always hooked on the app by showing things that will earn your interest in looking at the screen. This is only possible because the machine will be observing your behaviour, usage patterns, what you like and what you frequently look at.

Which means you are indirectly generating data for the machine to learn from you. So that it can show you the content that you like.

Have you, at any point of time experienced, when you browse on Amazon for something specific, later when you open Youtube or any other web app you will see similar ads related to what you had browsed on Amazon. This is because your usage is being used as a dataset to train algorithms in the background to show you exactly what you need to see.

Let's try to understand what exactly is AI and what exactly is machine learning.

By definition AI means “the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings.”

Artificial intelligence as the name states is the intelligence of a non human. In this case a machine/ a computer. Which means algorithms running on a computer which can take decisions, mimic human actions or emotion is technically called an artificially intelligent machine/system.

The best example for this would be a self driving car, which has the capability of driving you from point A to point B without the need of a driver. The car acts accordingly to the real time situations,while travelling on the path from A to B.

So how does a machine acquire the intelligence to decide/ act according to the situation or to predict the future possibilities?

The answer to this question is machine learning. Machine learning is a way to implement AI.

Machine learning is a process of training a machine based on relevant data to perform intelligent tasks.

Machine learning is a subset of AI. Machine learning involves the following steps.

1. Identifying the use case.

2. Gathering the data.

3. Processing/formatting the data.

4. Creating datasets.

5. Choose an appropriate model/algorithm to train.

6. Train the model.

To relate to this, let’s get back to the self driving car example.

Imagine yourself driving a car and the senses that would be in use are :

1. Eyesight – To understand what is in front of you or behind you.

2. Hearing – To listen to all the sounds that are around you.

Taking inputs from these sensors, your brain will judge and send signals to your motor functions to act accordingly.

Similarly for a machine to drive a car it needs data from its surrounding, which is possible with the help of sensors like camera, ultrasound sensors, RADAR, LIDAR and IMU sensors etc. Just like a human mind, the processing unit gathers data from all these sensors and make judgements and then it sends signals to actuators which performs actions similar to a human.

This is the basic understanding of how a machine can drive a car without human intervention

Chapter 3: Man and machine.

If we compare a man and a machine, they are not so different. The end goal of super AI research is to build and train a machine which can think and act similar to a human. In fact it can be much more intellectually superior than a human.

"We are living in a simulation" ~ Elon Musk

The simulation theory has been suggested by Elon and many other people. If we take a look at the computer games of this current generation and compare it with the games of the 1990s, we have come a long way. Games of this generation look very realistic and are very immersive. With technologies like Virtual reality, the immersion is greater where you can experience existing in a virtual world.

In the days to come, the reality factor can only be improved. In the next 10 years or so the games and experiences can be more realistic where one cannot differentiate between the real and virtual world.

Even if we believe we are living in a simulation, what might be the purpose of this simulation?

Theory 1 : Humans are a dataset

7.8 billion in the number of people living on Earth. If the entire population of Earth is considered as a dataset, it's huge.

Earth is a virtual environment, where data samples such as us are living our lives. Each data sample(person) is unique but there are certain properties which are similar.

Relating back to machine learning. In machine learning, a feature is a single measurable characteristic property of the data sample being observed.

Similarly if u consider a human as a data sample, humans have a standard set of features(properties) which can describe them.

Let's take a look at all the basic features(properties) of a human:

Physical features :

1. Height.

2. Weight.

3. Skin complexion.

4. Color of hair.

5. Eye color.

Non physical features :

1. Speech/Language.

2. Skills.

Emotional features :

1. Empathy.

2. Sorrow.

3. Anger.

4. Happiness.

5. Fear.

6. Surprise.

7. Disgust.

Other features :

1. Curiosity

2. Determination.

3. Passion.

4. Grasping.

These are just to name a few features of a human. Even though we may have missed some features, end of the day it's conclusive that humans have a finite set of features that can describe them.

Even though all humans may have these features, they vary in proportions in each and every person. Initially in a data sample's(person's) life, these proportions are hard coded into their DNA, as the genes are being passed on from their ancestors. Later these proportions may increase or decrease based on the experiences that that data sample may come across.

Example 1 : Imagine a person being very scared of heights, the features in play here are fear, surprise etc. There comes an opportunity for him to skydive and he is forced to skydive. Now the proportions of the features may change.

Before skydiving:

1. Fear : 80

2. Curiosity : 3

3. Courage : 0

4. Determination : 0

5. Surprise :70

After skydiving:

1. Fear : 20

2. Curiosity : 30

3. Courage : 35

4. Determination : 70

5. Surprise : 20

In this case skydiving was an event which had direct influence on the data sample's features.

So the next time he skydives his features/ emotions will be totally different.

Example 2 : Imagine a classroom of 50 students. Each student will possess varying feature proportions. So some samples among the 50 dataset(students) will have good value of intellect for subject A, and another set of samples(students) will have a good value of intellect for a different subject, say subject B.

This is the reason why there won’t be uniform results when you test the data samples(students) on a whole. Some a excel both in subjects A and some may not excel in both subjects A and B

This proves that every human if considered as a data sample will possess unique features which describes them.

There is another concept in machine learning called “Reinforcement learning”.

Reinforcement learning is a process where the machine must explore unexplored options or exploit current knowledge in order to increase cumulative rewards(rewards being nothing more than a mere numerical value).

In simple terms the machine gets rewarded if it performs expected actions and will be punished otherwise.

Now let’s relate this to human beings.

The brain is the most important part of a human, any other part of a human can be bionic including the heart. But the brain cannot be replaced(at least not at the time this was being written).

The human brain releases a number of chemicals and also is responsible for reflexive actions and pain/pleasure sensation. The main purpose for this is, it helps a human survive. In accordance to Darwin’s theory “Survival of the fittest”, the human brain acts as a tool which helps the individual to survive his environment.

What might be the purpose of surviving?

If we look from a broader perspective, the human body especially the brain seems to be designed to learn in a reward based learning environment.

Starting from school, students are rewarded with grades for their academic tasks. Working professionals get promotions or hike in the salaries.

If we step down to the biological level, the brain secretes a chemical called “Oxytocin“ when you procreate or create a social bonding or anything that gives you a sense of pleasure.

In a way the brain rewards you with oxytocin when you do something that you intend to do.

The other way is how brain percepts pain. The human brain is a tool to detect pain in any part of your body. This is another survival measure taken biologically. The purpose would be to avoid things that can be harmful for your survival.

Relatively, if someone is building a machine that can act, thinks, talks and walks like a human, the builder must also be intelligent enough to provide the machine with the intelligence to ensure its survival in its environment.

An epoch is a term used in machine learning that indicates the number of passes of the entire training dataset the machine learning algorithm has completed. The target with each epoch is to increase the accuracy/intelligence of the model.

Comparatively, if we look at humans as a dataset, through each century(epoch) that has passed since the beginning of time, the intelligence of humankind as a whole has always been an increasing trend. With the passing of each century, a new discovery is made or something spectacular is invented which marks an achievement of the time.

So in a way, both humans and machines learn based on experiences. And sometimes learn in a reward based environment.

Chapter 4: Evolution and boundaries.

Life began on Earth as a single celled organism, which kept on evolving until this point of time, into millions of species. Life kept on evolving, adapting to the environment.

Studies show man evolved from a more primitive being like an ape. If we can relate this to the digital age, try to remember how the first computer operating systems were VS how the current operating systems are. Another example would be the first ever smartphone VS today's most advanced smartphone.

In both these examples, what started out as a primitive tool, evolved into a technological marvel. This was possible because regular iterations of updates. Just like how initial versions of a human had canine teeth and now they don't. The explanation is, canine teeth which are used to tear into flesh or chew on raw food is no longer required by the modern version of a man. These information are passed on from the genes over time, so that from one generation to another, life evolves in accordance to adapt to the environment around them.

Evolution can be compared to the bug fixes that you get along with the software updates on your smartphone.

Chapter 5 coming soon.....