‘I think, therefore I am’. Is it possible for the Artificial Intelligence to repeat Descartes’ famous dictum and declare its intellectual independence from human control and become a thinking digital object that possesses the highest specifications of rational simulation?
Such are open questions raised within the scientific milieus.
Before, it was an hypothesis brought to us by science fiction movies but today they are coming close to its realisation by approaching general intelligence (AGI).
We have already begun to feed some of these features of rationality through the latest models of AI such as ChatGPT, particularly its fourth version.
Before knowing the work mechanism of the digital brain, I would like to repeat a piece of information which is rife in most of the literature on artificial intelligence.
It is that the mechanism of artificial intelligence is inspired by the biological brain of humans even though the functioning of artificial neural networks is somewhat similar to the working mechanism of the human brain which was the origin before the emergence of the digital brain.
However, a number of biological details related to the functioning of the human brain are still shrouded in mystery.
What we ought to know from this article is the relationship of the functioning of both the biological brain and the digital brain in order to bring the idea of digital thinking closer to artificial intelligence.
Artificial intelligence has several methods. the most important of which is called ‘machine learning’ from which one of the most powerful and complex types of artificial intelligence known as deep learning, is derived.
Among the most efficient types of deep learning is the Recurrent Neural Networks mostly used for recognising and processing written texts. It is one of the main types that function with the generative conversational model ChatGPT.
The method of deep learning in artificial intelligence functions on the principle of learning or training on a set of data. The greater the number and quality of data, the greater the speed and quality of learning.
This type of learning is called the deep type and it works according to multiple mathematical algorithms that enables this digital object to refer to the general ability i.e its success in recognising entirely new inputs and its ability to deal effectively.
This digital object comprises three main layers: the input layer (data enters this layer to form a set of digital neurons), the hidden layer ( it can be a multiplicity of hidden layers with a lot of neurons depending on the degree of complexity of the neural model of the digital object) and the third layer is the output which shows the outcome from analysing the input. What we should care about is learning to relate inputs and outputs and identifying the mechanism that falls within the scope of algorithms.
The functioning of the digital brain and its superiority has something to do with the type of the algorithm and its working mechanism.
This algorithm depends on a mathematical learning mechanism that allows the modification of weights.
One of the most popular types of learning algorithms is known as Back-Propagation which is the main and most important factor in the deep learning system of artificial intelligence.
Oman Observer is now on the WhatsApp channel. Click here
