AI is Augmentation of Intelligence

Why AI is just the next step in an age old human quest

I have been talking and writing about AI for quite a few years now. Over time, I have formed my own idea about what artificial intelligence actually is. However, I have never written on this very topic. This is that article that I should have written a long time back.

In this article I put forward the view that AI is just the next step in the human efforts for augmenting their capabilities. In fact, it is the very next step after the emergence of the IT systems. In light of this, I explain how AI differs from traditional IT.

The (Unexpected) Dominance of Homo Sapiens

Imagine yourself somewhere in the east of Africa, 50,000 years ago. A competition for ‘the most promising species’ is being held. Human beings will not even make it to the qualifiers.

The big cats — lions, tigers, jaguars and cheetahs will stake a claim for the coveted prize with their strength, speed and hunting abilities. Elephants not only have the right looks for the award, but also possess high intelligence and a very versatile trunk. The Hippo can turn out to be a surprise winner. After all, it can occupy both the land and the water, has a fierce bite and can run at a scary 30 km/h. Humans will be considered too weak, too slow and too delicate.

Homo sapiens might not have won any awards in the ancient times, but they have emerged as the winners of our planet. More than seven billion members of our species now inhabit the earth. Humans have occupied all continents, including a temporary residence in the frozen Antarctica. Human beings and their cattle weigh around 97% of the weight of all mammals living. The other promising species of old are now either extinct, on the verge of being so, or are restricted to a small region.

What is the reason behind this incredible success? How come humans rose to such prominence while other capable species could not?

Humans are Compulsive Augmenters

Scientists, researchers and thinkers have been working on the question of why humans have risen to such dominance. They propose human intelligence as the main reason.

Ours is not the only species to possess intelligence, however. In fact, every living being has enough intelligence to survive, and some animals such as the primates, the elephants and the dolphins possess substantial amounts of it. But these animals don’t seem to have used their intelligence to compete with humans or with other animals.

It seems to me that the most important aspect behind human progress is not intelligence in itself, but the way it has been used by us to augment our capabilities. Augmentation means addition, and we have been continuously adding to our capabilities for the past thousands of years using our intelligence.

‘Humans dominated because they used their intelligence to augment their capabilities.’

Human beings are delicate by animal standards. For instance, they can only live in a climate that is not too cold, due to their lack of fur. But they overcame this with a few augmentations, that of clothes, shelter and controlled fire. Humans originally lived in warm North Africa. But these augmentations allowed them to explore the much colder climate of Europe. Humans used language, an aspect of intelligence, to collaborate and work in teams, which made them much more effective than the animals that usually act alone.

The next big step in augmentation was making tools. Humans started making tools very early. These early tools were very simple, but they proved very useful in overcoming the limitations of our early ancestors. They used fishing hooks and pointed spears for hunting, needles for sewing and knives for cutting. The augmentation of capabilities using tools thus got started.

Image by Rajashree Rajadhyax

As years rolled by, humans kept making more and more tools. Now they were called machines. Humans made machines for farming, for transport, for cleaning and even for making other machines. These machines gave the humans a tremendous advantage over their rival species. There was no way a hippo could now compete with the ships and cars that humans built.

Humans then started working on the augmentation of intelligence itself.

They created machines that could do calculations and named them as computers. Memory, another aspect of intelligence, was also added to the computers. Large databases and blazingly fast computation created such powerful intelligent machines the likes of which the world had never seen. This revolution was called the ‘Information Age’. The machines in the information age combine the intelligence of computers with other capabilities such as communications, sensors, controls and interaction with humans. Mobile phones, internet, e-commerce, digital finance are all examples of these machines that we now call ‘IT (information technology) systems’. By the beginning of the third millennium, IT systems became an integral part of human life.

But Homo sapiens are compulsive augmenters. They are not happy with whatever they have achieved with IT systems. They are already working on the next level of augmentation of intelligence which they call ‘Artificial Intelligence’ or AI.

‘AI is the next level of the augmentation of intelligence’.

Is AI different from IT? The technology industry has a liking for fancy names. In fact it keeps repackaging old things with new names. So is AI a new name for some old technology? Let’s discuss this in the next section.

IT v/s AI

At the heart of all IT systems is a computer. A computer does its job by running a program, which is nothing but a set of step-by-step instructions.

See the following simple example of instructions to calculate the average of numbers:

1. Start with Total = 0.
2. Take one number and add it to Total
3. Repeat Step 2 for all the numbers you have
4. Divide Total by the count of numbers. This is the average of given numbers.

The programming languages such as C and Python help us to ‘feed’ these instructions to the computer. Computers are very good at running the instructions at tremendous speed. As an example, using the above instructions a normal computer such as your laptop can calculate the average height of 70 crore Indian males in about 10 minutes.

This method of solving a problem by step-by-step instructions is called the method of ‘algorithms’. Algorithms have existed far longer than computers. They have been used in mathematics for solving various problems since ages. But the invention of computers unleashed their power like never before. The programs on your laptop, the applications on the web and the apps on your mobile, are all based on algorithms implemented using various programming tools.

The method of algorithm is a big success, as we can see from the way IT systems touch all aspects of our life. If so, why are humans looking for the next level of intelligence? Is there something beyond algorithms?

There is. In fact, it is something that we humans are really good at. We do it at every moment in our life. It is: ‘guessing’.

Why do we have to guess all the time? Most of the problems that we face in our life have no firm answers. There are no step-by-step, well defined methods to reach solutions. Take for example the following problems:

• How much time will it take to reach my college or office?
• Is this person the right one for this job?
• Should I study further or start working after my graduation?

Our method of solving such problems is to make guesses. The guesses we make are based on a lot of input and thinking. But they are still guesses.

Algorithms are not good at guessing. They need well defined instructions and clear inputs to produce their results. The entire bulk of IT applications is thus useless in solving the problems that require guessing.

We will take an example that will further clarify this point.

Mangalyaan and Mangal Karyalaya

In 2013, India launched a mission to send a spaceship called Mangalyaan to the planet mars. The challenge was: how to use as little fuel as possible, as Mars is very far from the Earth? The answer was a very complex maneuver. Mangalyaan was first put into earth’s orbit. Its orbit was then increased in a series of steps. When the orbit was big enough to be close to Mars, it was changed in such a way that the spaceship started orbiting around mars instead of earth. Thereafter, the orbit was progressively reduced to bring Mangalyaan close to the red planet.

As you would imagine, working out all this in advance involved incredibly complex calculations. Some very brilliant scientists and engineers created the instructions. They were executed on truly powerful computers. It was a super complex problem and the solution was amazingly innovative, but the method was still well defined. That’s why it was possible to solve it using algorithms.

Now think of the following, much more familiar problem:

How much time will it take for your family to drive to a wedding venue (called Mangal Karyalaya in some parts of India), 10 km away from your house?

There is no definite way to calculate this. If you assume the car’s average speed is 40 km/hr, it will take 15 minutes. But the average speed depends on the traffic on the way. And the traffic depends on the time of the day, whether it is a weekday or weekend and so on. Even on the same days, traffic is not the same. You will need some current information about the traffic to estimate the time you will take.

If you notice, we are no longer trying ‘to calculate’. We are using the word ‘estimate’, which is another way of saying that we are guessing. We can imagine ourselves saying things like ‘the traffic on M.G. Road seems 20% more, so maybe our speed will reduce by 25%’. There are a lot of factors, a lot of approximations in solving this problem. In guessing, there is no guarantee of correctness. In fact, you don’t aim for the perfect answer. You aim for an answer that is good enough. In this case, you look for an estimate that will ensure you reach the wedding in time for the ‘muhurat’.

In his seminal book ‘Probably Approximately Correct’ (lovingly called the PAC), Turing Award winner Leslie Valiant calls such problems as ‘theoryless’ problems. We will simply call them ‘guessing’ problems. We will assign the name ‘compute’ problems to the classical problems that can be solved by algorithms.

AI is Guessing

Apart from reaching wedding venues on time, how are guessing problems important? How can they help us in industry and daily life?

Computers (and machines in general) have automated a lot of our tasks, but tasks such as prediction, decision making, design and generation still require human intelligence. We would like them to be automated as well, so that they can be done on a large scale. Let’s see some practical examples:

• (Prediction) A large departmental store would like to know how many of each item will be sold next week. They can place orders accordingly.
• (Recommendation) A songs streaming service wants to recommend songs to its users. The recommendations should take into account the users’ preferences.
• (Decision making) A bank would like to automate its loan approval process to handle larger volumes.

These are only a few examples from a growing list of tasks that we would like machines to do. If machines have to help humans in such tasks, they must develop the ability to guess.

In fact, we can now define the term ‘Artificial Intelligence’ better. Since computers are intelligent and obviously artificial, the term artificial intelligence should apply to all IT systems. But we can narrow it down using our new classification of problems.

AI is the ability of machines (computers, mostly) to guess, not just compute.

I hope that this is enough to convince you that AI is really a new thing and not just a new name for an old technology. The industry has found a new method and that’s why it is bristling with new energy.

Summary

There are two important takeaways from this article:

1. Artificial intelligence is the next step in pursuit of the augmentation of intelligence.
2. AI involves solving problems by guessing, rather than the traditional IT method of computation.

We also learned that there are important problems such as prediction and decision making that require guessing. Machine Learning (ML) is a method of dealing with such guessing problems. While studying ML, you should keep this at the back of your mind.