Deep blue

Introduction

Deep Blue is a chess computer (Computer Game AI) that beat a human player for the first time in a chess world championship tournament in 1997. It is based on the Brute Force Method or A-strategy. After the victory over world chess champion Garri Kasparov, a broad public discourse on the effects of artificial intelligence ensued. In addition, the question as to what extent Deep Blue and other chess computers can be considered intelligent arose for research.

This whole article is based on the text by Martina Heßler: „Der Erfolg der ‚Dummheit‘“, 2017.

Main Part

Initially, research on AI was intended to gain new insights into the human brain and human thinking. It should thus be possible to understand and map thought patterns. To achieve this goal, two strategies were possible: The A strategy, which relies on high computing power, and the B strategy, whose strength is selective problem solving, but its implementation is significantly more complex. In the case of Deep Blue - and the research of the following years - the A strategy, or Brute Force method, was chosen. This is a mechanical procedure that calculates all possible moves in a game of chess without distinguishing between useful and meaningless move options. This procedure "is based on a search algorithm, an evaluation function of the moves and a huge database with played games"(Heßler: p.11).

During their development, it was not intended that chess computers like Deep Blue would deal exclusively with chess. Chess was indeed considered THE characteristic of human intelligence, which is why the programmers dealt with the game in the first place. However, the shift to optimizing chess skills, rather than gaining knowledge through chess, met with criticism in the research field. Deep Blue had lost its original purpose - even though, by defeating Kasparov, it crossed a threshold of machine intelligence in the history of computer research.

Just like the question relating to / concerning the purpose of concentrating on chess, the question arises as to whether one can actually speak of intelligence in Deep Blue or whether one can represent human thinking with machines at all.

Hubert Dreyfuß supported/represented the view that some characteristics of human thinking could not be simulated by machines: for example, a computer was not capable of understanding language in all its implicit and performative dimensions, of distinguishing the important from the unimportant or of coping with the complexity of everyday human life. Instead of being able to master everyday life, Deep Blue, for example, is only capable of playing chess - very well, but nothing beyond that. In terms of chess, in which Deep Blue achieved outstanding results, it lacked essential human characteristics: "imagination, ability to combine, acumen, wit, courage, caution". (Geppert: p. 41 f.) Basically, the computer lacked emotion and gut feeling. According to Hessler, the abstract, logical thinking of a computer cannot be equated with human intelligence. In addition, the machine lacks reflexive abilities. It cannot justify moves that have been made in the game that are not comprehensible to humans. Based on this argumentation, there is the thesis that Deep Blue is stupid despite its performance: It is not actual intelligence, but behaviour that merely appears intelligent.

Through the victory over the chess world champion, Deep Blue was briefly attributed a kind of intelligence, which however, is revised again almost at the same moment. In addition, the game of chess is suffering a loss of reputation and is no longer considered an intelligent game. The moment in which a chess computer could have been considered intelligent is nullified by the change in parameters.

The alternative method of the B strategy might have come closer to the concept of human thinking. This is a selective strategy in which senseless moves are first recognised and the next move is decided on a situational basis. The similarity to human thinking lies not only in the networked structures but also in the selective way of working.

The next generation after Deep Blue was Watson, which made a famous appearance in the quiz show "Jeopardy!" in 2001. Watson sees itself as a learning system, owns databases and searches thousands of documents to analyse unstructured data. This leads to an understanding of ambiguity and thus helps it to correctly answer trick questions in the quiz show. Compared to its predecessor, it succeeds in recognizing and interpreting contexts on a statistical basis.

The questions that the text finally raises deal with the possibilities of interaction between humans and AI, as well as whether intelligence is the only means to successfully win a chess game or whether pure logic and computing power are sufficient. Hessler remarked that the moves that seem "stupid" are often the ones that lead to success.

Conclusion

Despite clear objectives, AI research ended up in a dead end. Instead of devoting its resources to gaining knowledge about the human brain, research drifted into the field of computer applications in chess and industry. Nevertheless, it did raise questions about the concept of human, artificial and machine intelligence. Deep Blue was regarded as intelligent because of its chess skills, but at the same time it also distanced itself from them. Furthermore, it and its history caused discussions in media philosophy as to whether artificial intelligence is or will ever be possible at all. Even if research did not follow its original path, it was nevertheless ground-breaking for today's (further) development in the field of computer research and artificial intelligence.

Deep blue

From GameLabWiki

Contents

Introduction

Main Part

Conclusion

Related Links/Research