Each encourages artificial intelligence. Technologies and machine learning techniques have improved. However, in this early stage, in its development, we may need to somewhat need to stop our motivation.

AA can already be seen in a large range of trade including cost marketing and sales, business operations, insurance, banking and finance. Shortly, analysis of human investment management and recruitment and maximum performance of people is an ideal way of performing a wide range of business activities. Its capacity runs in relation to the structure of the entire business environment. It is obviously more likely that AI value in the entire economy could be worth a trillion dollar.

Sometimes we can forget that AI is still a process in development. For its initial reason, there are still limitations to the technology that exist before us in the brave new world.

In a published recent podcast of the McNean Global Institute, a company that analyzes the global economy, the company’s head Michael Chauey, and director James Stika, discussed what the AAA limits are and What is going to be done.

The factors that are limited to AI are limited

Manyika said that AI limits are “pure technical”. They explain them to identify how the algorithms are doing. This is the result, the results and predictions what it does? Then there are practical limitations in data usage as well.

He explained that in the learning process, we are not only giving data to the computer to these programs, but also train them. “We are teaching them,” he said. They are trained to be labeled as a data label. Teaching a machine to identify objects in a photo or recognizing a variable in the data stream shows that a machine is going to break, feed them a lot of more labeled data It is indicated that there is a machine in this batch of data. This collection of data is not about breaking the machine and if a machine is about to break the machine out of data.

Chei has identified five boundaries, which is necessary to overcome. He explained that now humans are labeling data. For example, people are going through the traffic photos and can use automated driving cars to create the algorithm that needs cars to run to detect cars and LAN markers to create label data To apply

Manyika said that he knows students who go to a public library in label art so that the algorithm can be created. For example, in the UK, people of the group are identifying images of different species of dogs, using the labeled data that is used to generate algorithms so that computers can identify data and identify What is this?

He said that this procedure is being used for medical purposes. People are labeling photos of different types of tumors so that when a computer scans them, they can understand what the tumor is and what type of tumor it is.

The problem is that most of the data is needed to teach the computer. The challenge is to create a way to speed up the computer’s computer data.

Devices that are now used to use are included Networks Network (GAN). The tools use two networks – one of the right things creates and the second difference is that the computer is producing the right thing. Two networks compete against each other that allows the computer to do the right thing. This technique creates a computer in the form of a specific artist, or produces architectural style of other things, which has been observed.

Many have indicated that people are currently using other techniques of machine learning. For example, researchers in Microsoft Research Labs are developing in stream labeling, labeling data using this process. In other words, computer is trying to interpret data based on being used. Although stream is close to labeling, although it has taken major steps recently. Yet, accordingly, labeling data is a limit that requires further development.

There is not enough data for the AI. To counter the problem, companies producing companies are getting data for many years. Try to cut and cut the amount of time to collect data, companies are converting into a productive environment. Creating a synthetic environment inside the computer allows you to run more tests so that computers can learn more and more.

After that, the problem is to explain what the computer has decided to do. Known as a description, there is a case with the rules that can investigate an algorithm’s decision. For example, if someone is not permitted to be imprisoned on the bond and not anyone else, then someone wants to know why. Anyone can try to decide, but it will certainly be difficult.