Nowadays Artificial Intelligence, Machine Learning, Deep Learning becoming hot topics of discussion. Also, these are the main topics about which people have a dangerous misconception. I do answer questions related to these topics and I see that people have a big misconception about these topics.
First, come to the first topic AI or Artificial Intelligence. Everywhere companies talking about AI. Google’s CEO giving a speech on AI, sometimes adobe and sometimes intel sometimes apple. Nowadays Everybody talking about these topics. Famous philosophers say about it, some people warn about our future. Anyway, this is our future so far.
From where Artificial Intelligence idea came from? Let’s go to history. When people were making advanced machines, computers and things were growing so fast. Many experts believed that they can soon develop machines that can talk like us, think like us, and do like us. Many theories were created by the experts which said we will soon make such types of machines. Some scientists were against it. So, many controversies started. Experts wasted a big amount of money and effort but no result came out. After failing they ended the hype of AI.
But after some years this idea started again. The companies created a new word called Weak Artificial Intelligence or Applied Artificial Intelligence. Its background was from some companies but with time it was accepted by other companies.
How did the Weak Artificial Intelligence come? Basically, we broke the Artificial Intelligence into two parts, Weak Artificial Intelligence and Strong Artificial Intelligence. Strong AI is also called General Artificial Intelligence. But one thing you must remember that we didn’t achieve Strong Artificial Intelligence yet. If we get success to achieve this part then we will be able to create robots whose brain is like us and also we will be able to create virtual systems who will be intelligent like us. We only achieved applied artificial intelligence.
Difference between these two parts? There is one thing which is actually intelligent which satisfies the word ‘Intelligent’ but there is also another thing which we think intelligent but actually that is not intelligent. For example, We talk with Apple’s Siri, we think that Siri is very very intelligent. But if we talk with a human being then the human is more intelligent and creative in comparison with Siri. If we talk about the difference at the biological level, a big difference will come out. In today’s world still, the only human being can satisfy the word ‘intelligent’.
If you’re not concerned about AI safety, you should be. Vastly more risk than North Korea. pic.twitter.com/2z0tiid0lc
— Elon Musk (@elonmusk) August 12, 2017
Everything and every system which we called Artificial Intelligent are actually Weak Artificial Intelligence. They are not which were our aim. Some philosophers say that the word ‘Artificial Intelligence’ is spoiled by the companies for their promotion. When they promote products, if they combine the word AI with them it becomes more advanced to hear, people think it is a very advanced product. That’s why they made Weak AI and confused people.
In actuality Intelligence is a different word which we didn’t achieve on machines yet. Yes, many people say that we’ve figured out some of it but it will take time and in this century we will achieve the Real AI or General AI.
Weak Artificial Intelligence( Weak AI):
You talk with Siri, Google, Alexa, etc. these all are Weak AI. Also, some of you might have cameras that detect movements in your home and automatically warns you through email, face detection modules. All of these are Weak AI. When people see a machine exhibiting human nature, they think that the machine is the same as us.
Now, we’re gonna talk about Machine Learning and the differences between Machine Learning and AI.
Normally, when you do coding of a program, you set an input and a formula, when the input passes through the formula it gives an output. Again, if we talk about Bitcoin, you might know how Bitcoin moves up and down in Bitcoin Trading.
You know the basis on what Bitcoin moves, consider that you predicted for 5 minutes where Bitcoin will go. On which factor it depends? Some of you will say RSI(Relative Strength Index) or CSI or some of you will say Historical Graph Movements. Many people will say thousands of probable patterns. Any of them can happen if you see them in a different time frame.
Actually, we don’t know what can happen. Our brain can’t process all these things at one time. So, we sat down to create a program but we faced a big problem that we don’t know the mathematical formula !! so, we will take a hit and code and this code will do trials to create better code than him. It will do hit and trial using different combinations of given data. It will do and test millions of processes to create better code and will accept the results depending on their accuracy. After all these things maybe it is possible to get a program which can give 90% or more accurate results. It will be so complex that maybe we don’t understand.
Machine Learning VS Artificial Intelligence:
How ML(Machine Learning) and AI is different from each other. Remember one thing that in this section we’re talking about Weak AI.
Let’s take another example, a car drives automatically was a Weak AI. It had some fixed rules. It kept following these fixed rules. But when the creators shifted it from Weak AI to ML, it started improving itself automatically to a better form. Now it will learn from its experiences and according to the experiences, it will improve its codes. ML is a reliable way of achieving Artificial Intelligence. ML raises the accuracy of AI but it is quite different from AI.
How? if we see a thing for the first time, consider a person saw a laptop for the first time and if he sees another laptop that is quite different from the first in color, shape, etc. but he will figure out that it is a laptop. But if we introduce a laptop to a current system or robot-like Sophia, we have to process millions of photos of different types of laptops in his system. These systems are not bad because if we process thousands of photos of dogs if he sees a dog he will figure out it. But the negative point is, it requires a lot of data. It is an AI only for saying but in the background it is based on ML.
But if we were approached super intelligence, do you think we need to provide a large amount of data? we say that we will create superintelligence who have the same brain as us, then we don’t need to provide a large amount of data. Because of no requirement of a large amount of data, it will not be dependent on Machine Learning. This thing separates Machine Learning and Artificial Intelligence from each other.