The Chinese room is a famous thought experiment, proposed by John Searle in 1980, which refutes the statement of “Strong AI”:

The appropriately programmed computer with the right inputs and outputs would thereby have a mind in exactly the same sense human beings have minds.

Suppose English is the only language you understand, and you are sitting in a closed room with sufficient library of books that help you do English-Chinese translation and Chinese-English translation. Your task is to interact with a Chinese speaker outside room. The question is: If the Chinese speaker feels like she’s talking with another Chinese speaker in ths task, do you literally understand Chinese?

Chinese Room and Turing Test

You don’t need knowledge of Chinese language to simulate that you understand Chinese. Machines neither.

The task itself can be seen as a Turing test, and the design is similar to Von Neumann architecture. However, the Chinese room tries to testify the existence of “consciousness” or “understanding”, which is a little irrelevant to the statement of Turing. Passing Turing test doesn’t mean a machine has any level of consciousness.

Strong AI vs. Weak AI

Most AI researchers don’t really care about the strong AI research, at least before the mind and consciousness of human can be explained by physiologists.

There is not any formal statement of what is “real intelligence” yet. Deep neural networks have got higher accuracy than human in some pattern recognition tasks, but no one would say deep learning excedds human intelligence.