Searle’s Thought Experiment and Turing’s “Imitation Game”
Searle’s Thought Experiment and Turing’s “Imitation Game”
The development of the technology today has reached a very high level. In addition, the speed of development is constantly increasing and with every year, the capacity and productivity of computers grow. Today, it can be argued that humanity is on the verge of creating an artificial intelligence. There are many difficulties with determining the moment of creating artificial intelligence. The key theories that solve this problem is the Chinese room and the imitation Turing game. At first glance, they contradict. However, nowadays the development of artificial intelligence does not allow drawing conclusions about the adequacy of the imitation game of Turing according Searle’s experiment.
Essence and Principles of Searle’s Experiment
The current situation in cognitive science and, accordingly, the methodology of artificial intelligence in many ways resembles a humanitarian turn that has been realized in non-classical and post non-classical (modern) epistemology. In the theory of cognition, a turn was made to the problems of the complex structure of the subject to the study of the epistemic status of irrational components of cognition, the influence of the value-semantic sphere on cognition, the role of corporality in cognition, the study of linguistic, collective, social, activity, cultural and historical conditioning of cognition. Similarly, in the methodology of artificial intelligence, there is a turn that indicates quite significant shifts in the interpretation of the intellect and approaches to its modeling over a relatively short (historical scale) time. Currently, many researchers work on this issue (Cordio 53). At the same time, different approaches to interpreting the relationship between natural and artificial consciousness are applied in various directions of artificial intelligence research. In this regard, John Searle carried out his mind experiment in 1980 (Searle 1). He began researching the grounds of thinking about the relationship between the mentality and its carrier. He tried to think that mental states do not necessarily have to be brain states. This means that they can be implemented in any other medium, not necessarily that the creature is a human. It can be of non-biological origin. A creature can be arranged in a completely different way, for example as a computer. The main thing is that it functions in a certain way just as a person. If certain functional states are realized in this being, then one can say that he has mental states, and in some sense he has consciousness, a psyche. However, as a result, the scientist came to criticize this position.
In order to show the lack of conviction, he offers an argument as for the function of consciousness in a separate room. A person who does not know another language, for example Chinese, was placed in a separate room with cubes with hieroglyphics (Searle 2). A person does not know hieroglyphs; he has a usage instruction in response to certain requests coming from the outside. Then he looks at the instruction in his native language and manipulates these cubes; the answers are generally meaningful. However, the person in this room will never know what he said as he does not understand a foreign language (Cordio 53).
If this conclusion seems fair, then it shows that the existence of simple functioning is not enough to say that an object giving certain answers has some understanding of what consciousness and certain mental states are. In other words, the subject operates with signs at the level of syntax, adds them according to certain rules, but the semantic content of these signs is not available to him. It looks like the work of the computer without the existence of consciousness. All this leads to the conclusion that the functionalist program as a whole is unsatisfactory.
Thus, intentionality is a necessary property for the emergence of thinking, and this intentionality is the product of some intrinsic properties of the brain. A machine with the same properties is capable of thinking (Cordio 1). The ideas presented cannot be called complete (Cordio 1). No one has ever found the solution to important questions the answers to which should lie in the basis of thinking about consciousness and AI.
The Essence and Principles of the Imitation Game of Turing
Three decades before the publication of the experiment, Turing had formulated his famous criterion for a sensible machine, the Turing test (Muggleton 1). There are three basic versions of the Turing test two of which were proposed in the article “Computers and Mind” and the third variant is considered a standard interpretation (Turing 433). While there is some discussion whether the modern interpretation corresponds to what Turing described or it is the result of the misinterpretation of his work, all three versions are not considered equivalent, their strengths and weaknesses are different.
Turing described a simple game for parties, which includes a minimum of three players. Player 1 is male, player 2 is female, and player 3, who is the host of the conversation, can be of any gender (Turing 433). According to the rules of the game, player 3 does not see either 1 or 2 and can communicate with them only through written messages (Turing 433). Person 3 tries to investigate and guess the sex of another two players (Turing 434). The task of player 1 is to confuse player 3 so that he makes the wrong conclusion. At the same time, the aim of player 2 is to help player 3 to make the right decision (Turing 434). The first version is called Original Imitation Game Test, Turing suggests that the role of player 2 be played on a computer (Muggleton 2). Thus, the task of the computer is to pretend to be a woman to confuse the player 3 (Turing 433). However, the result of the game can change the evaluation of its success when the computer starts playing instead of one players (Turing 433). Turing proposed the second option. As in the “Initial Test”, the computer plays the role of Player 1 (Turing 433). The difference lies in the fact that both man and woman can perform the role of player 2 (Turing 433). In this version, both players 1 and 2 try to persuade the leader to make the wrong decision.
The main idea of the third option is that the purpose of the Turing test is not the answer to whether the machine can fool the presenter, but whether a machine can imitate a person or not. Despite the fact that there are disputes about whether this option was implied by Turing or not, the majority believe that this option was meant by Turing and, thus, combines the second option with the third one (Turing 433). This led to the fact that player 1 is a computer and player 2 is a person of any gender. Finally, it looks like following: the task of the experimentalist is not to determine the sex of players, but to guess who the computer is (Muggleton 1).
Summarizing Turing’s data, it can be stated that his thought experiment is an important tool for determining the inherent intelligence of a computer, but he does not give exhaustive information about the very method of its implementation. In addition, it does not answer whether the imitation of consciousness can be considered its duplicate.
Argumentation and Comparison of Searle and Turing
The existence of the concepts of the Chinese room and Turing test causes the desire to compare the results and the logical chain of experiments to draw conclusions about their adequacy. The ratio of the simulation game of Turing and the Chinese room in terms of assessing the presence of intelligence in a computer is very clear at first glance. The room is essentially an attempt to refute the theory of strong artificial intelligence. The simulation game is a proof that the theory of strong artificial intelligence is correct.
There is an opinion that the Turing test was sustained not by a person in a room, but by a system consisting of a room, a rulebook, and a person (Muggleton 1). Searle distracts attention from this fact hiding logical errors and concentrating on one of the components of the system and thus performing in this case a purely mechanical work. The system comprising of the rulebook, person, and room is reasonable and it understands the Chinese language. However, a counter-argument in favor of Searle’s theory is that one can force a person to remember the formal rules of answers for the system. According to the philosopher, the system, which now will consist only of a person, still cannot understand Chinese (Searle 2). That rejects the presence of the intellect in the subject of the passed Turing test.
Another version of the relationship between the Searle and Turing theory is that following simple instructions in the human language does not create an understanding, but this does not mean that such an understanding cannot be created at all. Therefore, it can be assumed that the effect of understanding is achievable if to create a robot with a self-learning program and equip it with sensors to perceive the world. Such a robot can interact with the world, cognize it, be able to learn like a child, and therefore begin to understand in the same way as a person. That testifies in favor of an imitative game. At the same time, summing up Searle’s information, one can object to the fact that in the case of a robot performance, the same argument can be applied. There is an opportunity to place a person inside the robot. Depending on what he saw through the hieroglyphs sensors, he would perform the motion of the robot’s body at the same time not having any understanding of what was happening, but only performed syntactic manipulations on the symbols.
As it is evident from the experiment, the objection to the Chinese room is based on the determination whether a person understands Chinese characters only by his behavior. If the computer behaves in the same way as a person who owns Chinese script, then it should be assumed that the computer understands it. This also speaks for the advantages of Turing’s imitation game (Muggleton 1).
It is possible to criticize the Chinese room because in its judgments about the terms “understanding” and “consciousness” Searle relies on intuition. However, many things are not intuitive even in modern science; thus, people should not rely on intuition in judgments about artificial intelligence. For example, aliens having a different form of consciousness come to Earth and cannot believe that people are able to understand anything since meat is inside their heads. In this situation, intuition lets down the aliens in the same way as it can let down in the case of the Chinese room. Searle simply does not accept the possibility of a strong AI insisting that it would miss a causal brainpower whose existence is not proven. Thus, he focuses on the difference between simulation and duplication. For example, no one confuses computer weather simulation with real weather. This argument speaks in support of the theory of the Chinese room, and we should not confuse the computer simulation of understanding with real understanding.
However, this argument is subject to criticism. It is impossible to distinguish the simulation from duplication. For example, it is difficult to say whether an artificial heart is a real heart simulation or a duplicate. The same applies to prostheses since it is easy to lean towards the opinion that people on prostheses walk but do not pretend to walk. Therefore, the principle of Searles simulation and duplication has serious problems and cannot refute the hypotheses revealed by Turing’s imitation game.
Summarizing the data obtained from the experiments of Searle and Turing, both experiments are very controversial. Discussions of the imitation game of Turing and the Chinese room have been going on for more than 30 years and no consensus has been reached. In many ways, the experiment of the Chinese room is criticized. It is characterized by some inconsistencies in logics and it is not completely rejected by scientists. This suggests that Searle’s theory cannot refute Turings argumentation. Even the very existence of the Chinese room does not refute the Turing test. Both these theories have their strengths and weaknesses; they are forced to exist in parallel until scientists gather enough information to win one of the theories, for example, the experiment. Thus, nowadays, it does not allow making conclusions about the adequacy of game.