Application of machine learning in Artificial Intelligence.
If I say, world is being revolutionized, would you like to believe it? Would you like to think that we are being unwittingly whirled by a gush called technological development? Though you don’t like to believe it or not, we should take into account that we are in the middle of a whirl of a rapid development of technology that has been speeding up throughout the last decades. Human being has invaded most of the areas of nature and his biggest step in this process is he has started outsmarting himself. He has started instilling his intelligence in inanimate machines. The concept of artificial intelligence is the quintessential result of this. But the fact which should be considered is, artificial intelligence is not either concept or fantasy or a scene that can be seen in a sci-fi movie but a reality. We are Humans are seeing new dimensions which have never been touched before.
But, what does the term artificial intelligence mean? There are various definitions for this term and simply it can be defined as the area of computer science, in which human try to make machines that can replicate human intelligence to do the tasks that are done by human. if Self driving cars, software that generate fake likes and subscribers for you tube channels and Facebook posts, speech recognition software, automatic medical diagnosis from radiology scans and etc.
How could machine do these tasks and how to train these machines to do these? How could programme a computer to do the tasks we want them to do? For this, algorithms can be written and according to those written algorithms, machines can be programmed. But these algorithms do not have a considerable capability to fulfill such tasks. Therefore, there is a new approach called machine learning, that changes the way of creating software that can give solutions for these problems. Usually, when an algorithm is used, machine is being told every detail of how to do the task that we want them to do. But, suppose that we want these machines to recognize the faces of particular faces of some people. In this case, we, as humans do not know how do even humans do this. We can recognize people specifically, by their special characteristics like long black hair, brown eyes, pointed nose and etc. But are they really specific for each of them? That is not true. Every human is a particular combination of some characteristics and human brain is itself a brilliant machine that could memorize each of these blends. But, creating algorithms for every single task is impossible. Machine learning is the mean that could make these machines do the things we want.
In machine learning, relatively simple algorithms are used and machines are taught by giving them large number of examples. Statistical algorithms are used by machines in machine learning to learn from examples. These examples are called data. Although these algorithms are very simple, they have an incredible performance when they are given enough amount of data. Deep learning or the deep neural networks are the most popular moments for this. Usually, as we all know, input of a machine is a bunch of numbers. This becomes very important and powerful because, machine learning can be applied for anything that can be represent as numbers. And when we have numbers, there are millions of ways to
There are several types of techniques that are being used in machine learning. When a huge amount of data in a wide range is given to these machines as input, sometimes machine learning can create new outputs. This is called generative model. When these machines trained a model on examples of input and output it is called supervised learning while when the learning algorithms are given a set of their own inputs to figure out the categorize, it is called unsupervised learning. When the machine gets rewards to the tasks it has done accurately, it is called reinforcement learning. No matter what the type of learning, they all use data and statistical algorithms.
Currently the world is excited about machine learning, because it can be seen in real world, other than in laboratories and in movies. As well as, the range of applications in which the machine learning can be used is huge. Machine learning is used in many fields of artificial intelligence. It is used in diagnosing disease, photos app of iPhone which allows the user to search photos for particular peoples, google search engine, automatic translation software, automatic assistance like Apple Siri or Amazon Alexa and etc. We are able to encounter machine learning, if we intend to do online shopping. Have you ever been surprised by the uncannily accurate product recommendations when you do online shopping? This feature of online shopping websites like Amazon, uses machine learning. These systems work effectively due to the massive amount of data with millions of their customers and users. Machine learning is used to build systems for people with disabilities. Moreover, machine learning is more accessible for people than programming. Therefore, if people could with a proper machine learning tool and a good data set, they don’t have to be programmers to contribute to artificial intelligence. It paves the way to build sophisticated systems according to their own will. Also, cancer diagnosis has become easier task because of machine learning. Other than these applications of machine learning, online fraud detecting, traffic prediction, natural language processing’s such as speech recognition, email and spam filtering, stock prediction, Banking Domain, prediction of potential heart failure, automating Employee Access Control are some other applications of machine learning. Social media platforms like Facebook and Instagram also use machine learning. As an example. In Facebook, it studies and records the activities, likes, comments and chats of the user and the time he spends on particular posts. Then it learns from your own experiences and makes page and friend suggestion for the user.
In everything in the world, there are two sides; good and bad. Machine learning also has good and bad effects. Many scientists have predicted about the dangerous outcomes of machine learning. It can be both marvelous and dangerous. People are afraid that, artificial intelligence will become more successful than human and one day it will take over the world. In the book called Super intelligence by Nick Bostrom he has given an idea about the potential of intelligence explosion. This means an artificial intelligence which is more intelligent than human will find another more intelligent artificial intelligence which in turn will invent more intelligent artificial intelligence and so on and finally, there will be a super artificial intelligence that will overpower the human race. This is just an idea, but we can’t ignore this supposing that it is impossible. Another most dangerous issue of machine learning is, if it could do the tasks that were done by humans previously, the jobs of people will be taken by machines. This has been clearly mentioned in the book, Rise of robots by Martin Ford. This can affect every kind of jobs in the society. Moreover, due to being controlled by a few people of same type, these models become biased. Thus, it will be a great threat to the whole human kind, if these systems will be gender biased or race biased.
But, as the ones who are responsible and the founders of artificial intelligence, human being should know its limits and threats as well as benefits. The sole purpose of the human should be serving and the making a society that is safe for humans to live. Thus, according to my point of view, humans should be responsible for the development of artificial intelligence, keeping in mind that, though it is being developed, the first and the basic purpose of developing and using artificial intelligence is to make a better world for the human kind.
0 Comments