What is Artificial Intelligence (AI)?
We can call Artificial Intelligence the software that enables a computer or a machine controlled by a computer to imitate intelligent creatures while doing a certain job.
Human history includes certain turning points. People who once made tools – tools with stones and sticks later learned to process metals and strengthened their civilization with much more robust tools – equipment. In this context, gaining mineral processing capability has been a turning point for humanity. In addition, we can count inventions such as writing, gunpowder, and the compass as turning points in the construction process of our civilization. Artificial Intelligence may be one of these breaking points.
How Does Artificial Intelligence Work?
The aim of Artificial Intelligence is to come up with algorithms that can learn independently and use what they learn to produce solutions to a problem. The most important feature that distinguishes artificial intelligence from other software is its independent learning feature. So how can machines learn? Let’s examine it.
What is Machine Learning?
We call this action machine learning, which is performed by Artificial Intelligence software that comprehensively analyzes the identified data, making new inferences and learning new things. Machine learning takes place in its simplest form: The machine with Artificial Intelligence determines the main similarities that stand out in the data with different approaches. Then, it learns and classifies the tags with these similarities by analyzing them. Machine learning algorithms are generally examined under two subheadings:
1. Supervised Learning
Learning algorithms that fall under this scope make inferences by using previously learned data and information.
In other words, the data and data sets to be used in the learning phase are predefined. With this predefined information, new situations are interpreted and inferences are made, new information is obtained and machine learning takes place. To give an example of the working logic of the supervised learning algorithm; Let’s say you will develop a system that can tell you whether the pictures you show are bananas or strawberries. For such a system, the supervised learning algorithm will do the trick. Namely: In order for the system to say that there is a banana in the pictures, it is necessary to define a data group with many banana pictures in the system.
So you need to teach the system what a banana is like. Then you should do the same with the strawberry pictures. We call this group of data entered to describe banana and strawberry as training data. We showed the data sets that teach the algorithm what bananas and strawberries are like. Now we’re going to show you again a new data set with pictures of bananas and strawberries. Pictures of bananas and strawberries will be mixed in this data set. But here we will not make a definition (teaching). In this step, we will ask the algorithm if there are pictures of bananas or strawberries in the final data group. The algorithm will determine whether there are banana and strawberry pictures in the last data set based on the banana and strawberry pictures that it was taught before.
2. Unsupervised Learning
Here, as in supervised learning, the algorithm has no predefined knowledge or experience. In other words, no teaching is done at the beginning. In unsupervised learning, a large amount of random data is shown to the machine. The machine classifies these data, which it has not encountered before, in terms of similarity and similarity of meaning, and these data heaps that have not been learned before are associated by the machine algorithm and learned autonomously. Let’s say you have a market on a busy street. Hundreds, sometimes thousands of customers come and go daily. You want to get a better customer experience and higher revenues by analyzing the behavior of customers coming to the market.
The artificial intelligence algorithm with unsupervised learning ability, which analyzes the data of all purchases made in your market so far, is likely to make inferences as follows after classifying the Coke sales: Cola sales increase by 70% on holidays. 60% of customers who bought coke also bought chips. 57% of the customers who bought coke also bought dried nuts. Existing coke stocks will be exhausted after 12 days according to the daily sales average. In the light of this new information provided by Artificial Intelligence, you can review the stocks in your market, zoom the aisles of products sold together and make long-term planning. After the necessary data is shown, it is quite possible for the unsupervised learning algorithm to learn new information and present this information by making such inferences.
Artificial Intelligence in Today
The usage area of Artificial Intelligence is expanding day by day. Artificial Intelligence, which is used in many sectors such as security, health, automotive, automation, information technologies and finance, provides great savings in costs and labor. Companies are aware of the power of Artificial Intelligence and what it provides. Investments made in this regard have now exceeded tens of billions of dollars. The financial value of the artificial intelligence market is expected to exceed $ 50 billion next year.
Usage Areas of Artificial Intelligence
To list some areas where this technology is used:
1. Image processing
We can also call this object recognition. Thanks to artificial intelligence, computers can identify living and non-living entities in an image. Today, vehicles that can take care of themselves without a driver thanks to artificial intelligence benefit from image processing while traveling. They act by analyzing the obstacles, traffic warning, and information signs, and the traffic situation.
2. Data analysis
Today, an enormous amount of data has been created. By analyzing these meaningless, giant data stacks with artificial intelligence, companies can determine new strategies and even get predictions for the future. For example, by analyzing the data of millions of people on e-commerce sites with Artificial Intelligence, highly effective and accurate advertising strategies can be created.
3. Voice Recognition
iPhone users know. There is a very respectable lady on phones with an iOS operating system. SIRI is actually an artificial intelligence product. Not only does it recognize the voices, but it can also even follow the instructions and do what is required. It can even chat. At this speed, it is not unlikely that robots will be produced that can befriended us in the near future. These were just a few examples. It is possible to reproduce these examples with many sectors.
Artificial Intelligence in the Future
When used correctly, artificial intelligence will improve our civilization as never before. However, future predictions and expectations about artificial intelligence differ. Some express their concerns by voicing the scenario of robots out of human control. Some say that this is an unfounded thought and that the robot is unlikely to have an independent will. True. It can be said that artificial intelligence on its own suddenly got angry and swore to destroy all humanity is an unwarranted conspiracy theory. But: What is known that a new Hitler will not emerge in the future and threaten all humanity by using Artificial Intelligence as a weapon?
We discovered atomic energy and learned to use it. But then? The world now is full of nuclear weapons. It is not even a problem that the world is destroyed by a spark. Artificial Intelligence is a very effective and useful discovery. It is up to humanity to continue to keep it useful and harmless. Yuval Noah Harari’s views on the subject are alarming. No need to be afraid of robots. We are the ones we should be afraid of people.