Artificial Intelligence, Machine Learning and Singularity in terms of Creativity – Ali Aksoy

I know the types who are “angry” when it comes to artificial intelligence or super robots. This state of being “irritable” is much more than fear about the possibility of a disastrous fate for humanity or a dystopian world. In fact, to me, it is almost a pathological matter for nature.

Because, on the basis of the main idea that artificial intelligence or super robots cannot replace humans, there is an idea that has been invented by religions for thousands of years. According to this; everything in the universe and even the universe itself was created for “man”. Although it may seem like the funniest claim in the history of mankind to those who have an idea about the size of the universe, this is a natural thing that is not even worth considering in terms of the “prejudices” inherited from his ancestors who thought that the thing called the universe consisted of the places where he was born and raised, surrounded by mountains that he saw more or less on the horizon. is the case.

This is the basic intellectual subconscious argument that underlies the fact that we make the world almost uninhabitable, and that we do things to other creatures that we would never want done to ourselves. It seems that if anything happens to us from the hands of artificial intelligence and super robots, it will come from the same logic. What one despises, perhaps, will eat his head…

It is not my intention to present a horror scenario. But I want artificial intelligence to never have a “creative” ability, and I want those who sing like a nightingale without any evidence to silence the deep voices coming from the hypothalamus and listen to this issue.

In order to discuss whether artificial intelligence can have a creative merit, we first need to quickly review what we mean when we say creativity, innovation, innovation.

When we say creativity, creative idea, what is usually meant, we are talking about something that is not yet in the same form, but is “new” in at least a few aspects. Okay, actually the merit of creativity should be something narrower, but what is meant in general usage is broader. In summary, it is new and different, different.

Besides being new and different, what is to be considered creative must be practically functional. Anything like a “magpie on the roof” is not seen as creative, even if it is new and different. Finally, despite having all these qualities, we call creative what we “like”, even if only relatively. Thus, we include artistic structures in the circle of creativity.

To summarize roughly;

  • New and Different

  • Functional (works)

  • Nice, liked by others

things i.e. ideas, artistic products, architecture, solutions etc. We call it creative.

For example, qualifying something as cheap, big or small does not make it creative. However, we can qualify a cheap, new, different, useful and popular solution as creative.

Now; After framing and defining the concept of creativity in this article; Now we can come to the debate whether artificial intelligence can be creative or not.

For this, first of all, we need to examine a “disease” (!) called “machine learning”.

If we define it roughly and briefly, machine learning; It is the name given to a software examining a large number of unspecified data (big data) according to millions of parameters and revealing the relationship between the data.

There is so much data that it is almost impossible for a person to know what is related to what in a jumble of numbers and information that seem meaningless at first glance.

Without further ado, let me give a simple example. When you advertise on a Google search engine to be at the top of any word search, Google’s machine learning software starts working and starts analyzing who can click on the word you entered when it is listed in the search results. This analysis includes all kinds of data, including “illegal” data such as millions of clicks, page visits, location information, written and drawn on social media, and even the topics mentioned in friend chats next to smartphones without infecting any digital media. done over.

Since such an analysis can be made, Google artificial intelligence performs a complete example of prophecy, achieving a very high clickthrough rate of 30%. That is, one out of every 3 people that Google shows your ad on the search network clicks on your ad. If you think about it a little, you can realize how high a rate this is.

Let’s say you own a restaurant. Hundreds of people come to your restaurant every day to eat. If you give good-detailed, high-resolution video images of the people who come to the restaurant, the order information of each person, the price paid to such a software (in fact, if you provide the identity information of the people who come to the restaurant, the result will not be eaten, but we will suffice for now) and all this data from the software If you want the software to analyze it, the software can make extremely accurate predictions about who will come in later on, what they would order, even if it was their first time. It makes this prediction by looking for relationships among many parameters that seem unrelated at first glance, such as people’s clothing, hair style, facial features, possible emotional states, time, weather conditions, and testing each of them. The more data there is, the more consistent associations it can make. Imagine for a moment that you are a seller and you know in advance whether the person you see will buy your product, if so, what price he will agree to, and furthermore under what other conditions he will be willing to buy! For this reason, we can say that in the near future, there will not be a person who is more “human wise” than a machine learning with enough data.

In terms of creativity, the reason why we mention the phenomenon called machine learning is that it can know the criterion of “likability, liking” that we have just mentioned above, with an accuracy far beyond our estimation and capacity. If you can load enough data into the machine learning system to know whether a photo will be liked by people or not, the software can make amazing predictions for you. Moreover, far from knowing whether it will be liked by a certain target audience in general, it can also give information about “who might be liked” even if it is a small group. All of this is a math, a calculation based on concrete data and finding out the relationships between the data. It’s supremely consistent because it’s based on statistics and math. If there is enough data, it can also predict and calculate the absurdities and contradictions that arise from the human being as probabilistic.

It can also make inferences about people’s relative and cyclical tastes if there is sufficient data on the past and present.

When a record player says “this song is worth it”, it processes ten hundred million times the parameters on which it is based, and makes connections between many seemingly unrelated things.

So, we now understand that when artificial intelligence has enough data, it can make inferences about whether what it produces will be liked and by whom and under what conditions. Because something a Japanese may like, an African or an Arctic person may not.

Contrary to what we think, no creative idea or work of art emerges without an example. That thing is necessarily based on something else that exists in its owner’s mind. He definitely leans on a presence. What makes it creative is to bring forth an unprecedented mix of existing things. The more new, different, useful and likable this work of mixing, associating or, on the contrary, sorting out, deciphering is, the more creative it becomes. Why did we say this, because just as we build something on existing ones, artificial intelligence will build on existing ones. In the name of creativity, expecting artificial intelligence to do something that we do not actually do can only be a result of the pathological situation I mentioned at the beginning of the article. If Mozart had never heard musical pieces in his life, could he have composed those compositions? The answer is clearly “never”.

If artificial intelligence has enough data, it can know whether what it produces is new and different from existing ones. By analyzing it down to the smallest detail.

If creativity is the search for a solution to a problem, finding a useful, useful method, if you tell him what the solution is “for now”, that is, when all the parameters of the problem can be digitized, he will probably offer you not one, maybe thousands of alternatives and solutions with benefit and harm analysis for each. . What he has done in chess and the game of Go, he will this time show on the entire digitally definable field.

The interesting thing is that the evolutionary process that created the group “machines cannot be creative like us, brother” and many of the mechanisms of this process, such as natural selection, trial and error, gene crossing, are simulated in computers in the experimental environment, are made by learning software, and are used in the possible solution of many problems. This method is already used today. In other words, at the moment, the software already runs evolutionary simulations conducted with the same evolution logic in very large set and multivariate problems, reveals millions of “baby” ideas, methods and possibilities, eliminates them according to predetermined success parameters, and produces successful features like itself. It helps scientists or other interested parties to find the best possible solution candidate and idea by blending them with the survivors and doing all these operations over and over again thousands, millions, billions of times in a very short time.

Yes, man is creative, but he simulates what he creates in his brain, envisions it, associates it with other things, and finally reveals what he thinks will be most liked. It does this not with a fictitious hide-and-seek player called the “soul”, which has never been detected, but with billions of electrically powered neuron cells with data and connections. All these neurons, like the rest of their body, operate according to the instructions of a basic software called DNA. It is magnificent, fragile, limited.

We are close to reaching all the secrets of neurons in a human brain.

When this process is completed, just as a tiny calculator can easily perform a mathematical operation that we cannot do, artificial intelligence; We will see in a very short time that he can do many things that we think are unique to human beings, such as creativity, problem solving, and searching for solutions, which are much more complex and that humans act based on software, in a very well thought, deeply analyzed and very admirable way. .

One day, artificial intelligence will call us just like the “dabbet ül ark” in the Qur’an and remind us that we are actually just a “nothing” in the universe!

Ali Aksoy – 03.09.2020

If you liked this article, we strongly recommend that you read the following articles by Ali Aksoy.

About the Concept of Singularity – What is Singularity Actually?

Devil and delusion in the age of singularity

Is there free will – 1: Proven Illusion

(Translated with google translate)

For the original article, see:

Yaratıcılık Açısından Yapay Zeka, Makine Ögrenmesi ve Tekillik – Ali Aksoy