Successful in old age: Late developer Artificial Intelligence!

Illustration and Text by Susanne Gold

Success with over fifty

Not everyone is born during a conference, but our artificial colleague is. Already in the last century he saw the light of day – at a college. Another special feature is connected with his birth: his mother is unknown! His father named him “Logic Theorist”. Why is it so late in success?

In 1956, the term “artificial intelligence” was coined during a conference at Dartmouth College in New Hampshire.

There, scientists argued that aspects of learning as well as other characteristics of human intelligence can be simulated by machines. The term “artificial intelligence” was proposed by the programmer John McCarthy.

The “Logic Theorist” developed there – managed to prove several dozens of mathematical doctrines – and is considered the first AI program in the world.

 

Man and his artificial counterpart

 
 

Smart systems can make human decisions through fast and comprehensive analyses. Artificial intelligence is now the position to make decisions which make human choices and already exceed it.

This is mainly due to the ability of the machines to scour and compare huge mountains of data, the so-called big data, in seconds. A work that would have required many hours of research for humans and yet would not have been able to compete with the artificial brain: this can be 70 trillion data points from the Internet along with associated and unmanageable networks. Open. 

As a result, smart machines are increasingly creating new freedom and opportunities for people. While collecting, processing and analyzing data was once time-consuming and resource-consuming, it can now be done easily and quickly by a machine.

For many companies, this means that tasks that were once done by humans in time-consuming work can be automated by algorithms. Ideally, the employees are used for the interpretation and implementation of the analysis results.

A few years ago, AI was still regarded as a niche scientific research and was primarily used as a material for science fiction literature. Today, technology has long been an integral part of everyday life.

 

Why this hype today – What happened?

 
 

Big data is the key word that has been on everyone’s mouth in recent years. This refers to the huge amounts of data produced by the spread of the Internet, social media, the growing number of machine sensors installed and the Internet of Things (machines communicate with other machines).

This includes, for example, customer and credit card data that is registered at the checkouts of department stores, as well as the analysis of unstructured data, for example in the form of annual reports, e-mails, web form free texts or customer surveys, which are often are part of analyses.

The amount of data available is available through the Internet of and through the use of mobile devices (in addition to PC, smartphone, tablet and other smart items, e.g. smart TV) in virtually every household Increased. Added to this is the high speed at which the data is collected today, can be processed and used.

Big data, cheaper computer technology and better algorithms are making AI a mass phenomenon. There are four developments that are driving and influencing the triumph of artificial intelligence.

The Oil of Our Time – “Big Data”

 
 

Today, an incredible amount of data is stored and evaluated. These huge amounts of data are also called “big data”. Companies face the challenge of storing and analyzing the amount of data that is generated both efficiently and effectively. Because the data can be used to gain insights that can be used profitably, the data is also referred to as “the oil of the 21st. Century”

 

The speed of data transmission

 
 

Today, data is generated at very high speed, stored, processed and analyzed. This is made possible by a much better hardware than before. With the help of so-called “memory technologies” this processing has become possible, but also through adapted software, i.e. algorithms.

 

The diversity of data

 
 

Big data provides a wide variety of data and presents AI with the task of analyzing not only structured data and tables, but also data from flow streams, images or videos. That is already 85% of the data volume. The reason for this is our social media posts. There is a huge amount of unstructured data, the meaning and content of which can be captured using AI technologies.

 

Security

 
 

The credibility of data is becoming increasingly important. With the increasing volume of data, the falsification of this data is also increasing. Around global data security is therefore considered a megatrend for our common future: cybersecurity!

The point is that not all stored data is credible and should be evaluated. Examples of this are tampered-on sensor data, phishing emails, or, at the latest, fake news since the last US presidential election. AI technologies are also used to distinguish real data from fake data.

Where our artificial colleagues press the school bench

 
 

Man and machine have one thing in common: they have to learn to know and master something. Machines train their knowledge using large data sets, the so-called training data.

Large US corporations in particular are investing heavily in the further development of artificial intelligence. To improve AI, you need large amounts of data to train it. Only today, with the improved technologies, are these mountains of data available for modeling and training the artificial intelligences.

 

Successes

 
 

The greatest successes of artificial intelligence have been achieved in recent years through so-called “deep learning” techniques. Before there was training data, learning successes of aI could only be achieved with great effort and expert knowledge. For this reason, machine learning has long been considered a niche research. Today, however, as large datasets can be used to train artificial neural networks, results are improving at a rapid pace.

As a rule of thumb, a supervised deep learning algorithm with approximately 5000 categorized samples per category and a training record of at least 10 million examples should be at least human-like or even better. Through the training possibilities, the topic of machine learning has developed into a top-level research and today influences all areas of our lives and creations.

Data in itself does not bring any value. It is only the algorithms that create value from data.

Since it has become possible to analyze large amounts of data quickly, it is also at the heart of today’s industry and its production. That’s why our artificial friend – at least over 50 – celebrates his first professional successes.

 

Leave a Reply