Artificial Intelligence

Artificial Intelligence (AI) is a part of computer science. With research and development of artificial intelligence, it is attempted to not only simulate but to supplement human thinking with self-learning computer programs. AI is already widely used in business, for example, in Google’s algorithm RankBrain. The terms “neural networks” and “deep learning” are closely linked to the development of artificial intelligence.

History

The history of artificial intelligence began in the mid-1950s in the US. At a scientific conference in Dartmouth, M. Minsky, J. McCarthy, A. Newell, and H. A. Simon were the first to speak of “artificial intelligence.” A later frequently cited definition for artificial intelligence was given by one of the founding fathers of the subject, Marvin Minsky, in 1966: “Artificial Intelligence is the science of making machines do things that would require intelligence if done by men.” Thus, it was determined that artificial intelligence is a science and secondly that machines can take over the work of human beings which requires human intelligence.

A first product of artificial intelligence was the general problem solver of the researchers Newell, Shaw, and Simon from the 1960s. The device could solve simple problems. However, the research results of the apparatus cannot be generalized. In the late 1960s, another program was written with ELIZA. In this case, Joseph Weizenbaum, a MIT researcher, simulated a therapy session.

In the following years the still young science was being constantly further developed, resulting with the MYCIN in the early 1970s in yet another innovative system based on AI. The MYCIN was able to assist physicians with diagnoses.

The advancement of systems with artificial intelligence has been fueled by the ever-improving memory capabilities and processor performance of computers. Another highlight is IBM’s “Deep Blue” supercomputer, developed in the 1990s. The system is no longer based solely on human input, but can also learn by itself. The computer was able to play a game of chess in 1997 with the then-world champion. After six games, the computer won.

With the growing importance of the Internet, the possibilities of applying AI have also grown. In 2016, Microsoft wanted to launch a Twitter account, which was based solely on artificial intelligence. At the same time, however, the system showed the limits of AI. After a short time, the account published only racist tweets and tweets which were directed against women. The learning performance of the account was probably manipulated by a coordinated action. Microsoft closed the account for the public after only one day.

Search engine giant Google caused a stir when an employee reported in October that Google was using artificial intelligence to answer search queries.[1]. Google calls its AI system “Rank Brain.” In March 2016, Google publicly announced that “Rank Brain” was one of the three most important ranking factors. [2]

The Turing test

The “Turing test” plays an important role in determining artificial intelligence. The test developed by the British mathematician Alan Turing in the early 1950s is based on the fact that a person communicates in parallel with a machine and other people via a kind of chat program.

The prerequisite is that there is no visual or audio contact. If the test persons after the “conversation” cannot determine exactly whether their conversation partner was a person or a machine, the machine is considered intelligent. To this day, scientists assume that today’s artificial intelligence will not be able to fully pass the test.

Background

The definition of artificial intelligence is based on the idea that human intelligence is the sum of different calculations. Thinking men have been regarded as machines since the Enlightenment.

Artificial intelligence itself is produced in different ways.  

  • Pattern recognition: AI systems recognize patterns and can perform appropriate actions.
  • Access to a large knowledge base: Some AI systems are fed with a lot of knowledge. The system draws on solutions or answers from this dataset when searching for answers or solutions.
  • Prediction of patterns: By calculating probabilities, certain AI systems can react to possible patterns in the future.

Overall, artificial intelligence today is based on the processing of very large amounts of data, called big data. The most modern form of artificial intelligence uses artificial neural networks and develops self-learning systems in the form of machine learning.

Areas of application

Today’s use of artificial intelligence is multifaceted. Companies now have the opportunity to maintain their customer relationship management with chatbots. Purchases or warehouse management can also be based on AI systems. In robotics, AI is applied in the production of devices or machines, while artificial intelligence is also used in driverless cars, for example.

In principle, any area dealing with digital data processing can benefit from artificial intelligence.

Limitations of artificial intelligence

As helpful as artificial intelligence may be, many dangers and problems can arise through AI as well. That is why researchers are continually trying to identify the risks of using self-learning systems.

In March 2016, Google researchers formulated specific questions about artificial intelligence.[3] They wanted to clarify possible safety risks of intelligent machines. For example, the scientists questioned how negative side effects of the work of machines can be eliminated. Moreover, the researchers wanted an answer to the question of how an AI can safely explore its environment without endangering people.

Examples of artificial intelligence

  • IBM Deep Blue: IBM’s supercomputer has already played several chess games against world champions.
  • Cortana: Microsoft’s voice control is based on AI.
  • Siri: Apple’s voice control can respond to questions and perform tasks for its users.
  • Echo: Amazon’s voice control can independently place orders.
  • DeepText: Facebook’s AI analyzes email and status updates.
  • Stock exchange calculator or stock reports: Programs are used to calculate the probabilities of exchange rates. Stock market reports are already being created by AI systems in some media houses.
  • Offices: In the USA, AI systems are already being utilized to search emails, letters and legal texts, in order to independently develop contracts.

Benefits for online marketing

There is so far no direct way for the public to use artificial intelligence for online marketing. However, artificial intelligence could be used in online marketing in the future, for example, to automate ads. The system would then independently recognize which ads were successful and which were not. Such a system could then adjust ad copy independently.

There are already numerous companies which utilize AI in [[E-Commerce[e-commerce]]. This includes the mail-order company Otto, which uses AI to calculate product sales forecasts. This company has been using AI since 2008. Likewise, one can assume that large online portals such as Amazon already rely on AI.

References

  1. Google turning lucrative web search over to AI machines bloomberg.com. Accessed on 08/05/2016
  2. Now we know: Here are Google’s top 3 search ranking factors searchengineland.com. Accessed on 08/05/2016
  3. Bringing Precision to the AI Safety Discussion research.googleblog.com. Accessed on 08/05/2016

Web Links

Category