In modern times, humans have been trying and dreaming about developing artificial intelligence. The story of developing artificial intelligence started in late 1950 when a machine was created that uses language that can fix problems and issues. Now humans are trying to improve this concept decade by decade. Officially, that time was marked as the beginning of the history of AI or artificial intelligence.
What Are AGI and Its Requirements?
It is challenging to define AGI or artificial general intelligence. It is because, “General” term is a broad concept, and if human intelligence is considered as the baseline, then all humans don’t have an equal intelligence level.
Yes, some traits support the intelligence system like background knowledge, common sense, abstraction, causality, and transfer of learning.
Main Approaches To AI
Symbolic AI and General AI
The development of the computer programming language was based on symbolic manipulation. But it has fundamental loopholes. Symbol AI only works as long as you try to encode the logic of the tasks into the rules. During the 1980s, the scientists working on AI tried this approach with their expert systems. The expert systems were successful at that time for narrowing the domains, but they failed in expanding their reach and addressing more general problems at the time.
For solving this issue, symbolic AI must add more rules that can compare the pixel of each new image that you have gathered.
It is a simple image of the basketball, but now imagine using a more complex object like a chair, or other deformable objects like shirts or a piece of cloth. The tasks will become pretty tricky for symbolic AI.
The history of human intelligence shows that symbolic manipulation is one of the several components of the general AI.
General AI vs. Machine Learning
Artificial intelligence develops its behavioral aspect through their experience. Deep learning is one of the crucial aspects of the branch of machine learning. In fact, this field is receiving a lot of attention in the past few times.
Neural networks are known for establishing and dealing with messy data like un-managed data such as audio files and photos. Even in some years, speech recognition and processing of the natural language have the part of advanced computer vision.
To return from the detection and identify the object that would be solved with the deep learning concept. For that purpose, you have to create a convent, which is the type of neutral-network. It is known for processing visual data.
In fact, instead of comparing the photograph from pixel-by-pixel. The development of the mathematical representations works with a deep neural networks model. Even if you compared this with the symbolic AI, it was found that neural networks are slightly better in identifying the objects in images.
When Will Artificial General Intelligence Become A Reality?
Many knowledgeable people are working on this project; even some of this project is under process for several decades. Presently, several researchers are aiming to generalize the capabilities of Artificial intelligence.
There are disagreements and arguments in developing a single direction regarding artificial general intelligence. There is a lot of research doing on developing a system with the deep leaning aspect that can help you in performing high-level symbolic manipulation.
These approaches eventually will bring you closer to the AGI, and they will also work in unveiling more loopholes and roadblocks. There is doubt in the above statement, but one thing is confirmed that there will be a lot of discovering along soon.