Recognizing Artificial Intelligence (AI) complete and recent-in the 1950 ‘s, Minsky and McCarthy, making Artificial Intelligence (AI)/artificial intelligence as a task performed by a program or machine where if people do the same activity, then humans must use their brainpower to accomplish the task.
Arguably the definition is quite broad, which is why sometimes we will find an argument about “whether the system is actually done by an AI or not”.
Get to know Artificial Intelligence (AI)
Artificial Intelligence (AI) systems usually have at least some similar behaviour to human intelligence, such as:
In the lower Artificial Intelligence (AI) system, there is also social intelligence and creativity.
Purpose of Artificial Intelligence (AI)
Nowadays Artificial Intelligence (AI) is already in the daily activities of human beings, perhaps we do not realize it. Like online stores that use Artificial Intelligence (AI) to recommend a product to us and to understand all our sayings when using Virtual assistant apps like Amazon Alexa and Apple Siri.
Then Artificial Intelligence (AI) is also made to recognise who and what’s in a photograph, find spam in an email, or detect credit card fraud.
Different types of Artificial Intelligence (AI)
Artificial Intelligence (AI)/artificial intelligence can be divided into two types, namely Narrow AI & General AI. Narrow AI is an artificial intelligence that is already used by humans, one of which is in the computer system today.
The Narrow AI system has been created in such a way as to be able to learn to perform certain tasks without having to be programmed further to be able to accomplish the task.
The real evidence we can see in speech recognition and language from virtual assistants like Siri on iPhone, then there are also cars that can move on their own without having to be controlled by humans, and lastly in online stores that suggest a product you might like based on your item purchase data history online.
Unlike humans, this system can only learn or be taught a way of doing certain tasks, then from this artificial intelligence system called the name Narrow AI.
The task that Narrow AI can do
There are a large number of applications that use Narrow AI. It is usually used to interpret a video from a drone that is conducting a visual inspection of a building, such as an oil pipe.
Detect damage to elevators based on data collected by IoT devices, etc.
Then create a custom work calendar for users of specific applications, respond to customer service requests, and coordinate with other systems to perform tasks such as booking hotels automatically at the time we have specified.
In the field of health, artificial intelligence can also help radiologists to find potential tumors in the results of X-ray photographs.
And in the Internet world, AI is usually used to block inappropriate content.
Tasks you can do with General AI
General AI is very different compared to Narrow AI. General AI is a type of artificial intelligence that can adapt. This is similar to human-owned intelligence.
Generai AI can also be said to be a flexible form of intelligence that can learn how to accomplish a job in a very different way. This system can be used for robots to cut human hair.
But, unfortunately at this time General AI can only be met in science fiction films. Like Skynet in the Terminator movie, but until now the experts are still unsure about how fast the General AI can evolve to have the same capabilities as Skynet’s artificial intelligence in the Terminator movie.
According to survey data in 2012 conducted by experts by Nick Bostrom and Vincent C Müller said that the probability of 50 percent of Artificial General Intelligence (AGI) will be able to be developed by humans between the year 2040 or 2050.
Furthermore, this group even predicts the “Super intelligence” will successfully be created about 30 years from the time General AI was successfully developed by humans.
For those who do not know about Super intelligence, Bostrom defines super intelligence as “any kind of intelligence that can greatly exceed human’s cognitive performance in almost any field we want”
But some artificial intelligence experts believe that the projection done by the group is very optimistic. Given our own understanding of the human brain is still limited, and they also believe that Artificial General Intelligence (AGI) still has to take centuries more to be developed by mankind.
What system is the machine learning?
There is a lot of research on AI, which most of these studies complement one another. Currently research on AI is undergoing a revival.
A machine is designed to know how the computer system enters data on a large scale and then the machine is used to learn how to understand human speech or to make a photo title.
AI’s Neural network
The key process of learning a machine is neural tissue. This network is a network that is inspired by the human brain and is made of interconnected algorithms. This network is called neurons.
Neurons give data to each other, and can be trained to perform certain tasks by modifying the important attributes of the data and then entering the data over between the algorithm layers.
The transmission of data through this neural network, will have a weighted data stick to different inputs, and will continue to vary until the output of the neural network is very close to the targeted thing, at this point the AI will ‘ learn ‘ how To perform a specific task.
The purpose of machine learning is deep learning, where the neural networks of AI are extended to other tissues with a large number of layers that are trained to transmit large amounts of data.
This network is a very powerful network that triggers the rapid development of computer capabilities to perform unique tasks, such as being able to recognize a sound input.
There are various types of neural networks of AI, of course with different strengths and weaknesses. One is a recurrent neural network, which is a type of tissue that is very suitable for language processing and voice recognition, while the convolutional neural network is often used for image recognition.
Neural network design has also undergone development. Recently researchers have perfected the form of a more effective deep neural network called Long Short-term memory (LSTM) in Bahasa Indonesia meaning short-term memory.
This network allows the AI to operate fast enough to be used in the system. Examples are like Google Translate.
Continuously developed AI research
Another field of AI research is computational evolution, where this calculation borrows the natural selection theory from Darwin. Where AI’s genetic algorithm was created to undergo mutations and to create random combinations between generations so that AI could develop the optimal solution for a given problem.
This approach has even been used to help design the latest AI models, where the AI can effectively help build other artificial intelligence.
The use of evolutionary algorithms used to optimize neural tissues is called neuroevolution. This algorithm has an important role in helping to design the creation of other more efficient artificial intelligence.
The new technique was exhibited by Uber AI Labs, who released papers on using genetic algorithms to train AI’s neural networks to conduct learning to solve problems.
Thanks to know Artificial Intelligence (AI) We can finally create a system where the computer is programmed in such a way as to allow the computer to take a series of decisions on its own based on a large number of inputs received .
Where it allows the computer to mimic human behaviour in a particular midwife. An example of this knowledge-based system, for example, an autopilot system that flies planes.