Teste de Turing e sua alusão a consciência de máquina

Abstract

From the initial studies on the competences of Artificial Intelligence, in the 50s, Turing, in his article Computing Machinery and intelligence, inserted the premise "Are machines capable of thinking?". Thus, chatbots are directly related to this idea, given that they are able to converse and learn, via deep learning, to adapt to different conversation scenarios. Therefore, the objective of this work is to compare the studies of the time with the current ones on artificial intelligences and to analyze if these intelligences are able to perform the “Turing Test”, na example of the results in the Loebner Prize, and if its accomplishment matches those pointed out by Turing. That said, at the end, after applying the test to the finalists of the Loebner award, the discrepancy in scores between champion Mitsuku and the other AIs can be seen. However, it was found that despite their high score, the problem of mimicry is still directly implicated as IAs who perform the test precisely with the objective of winning the test, thus interfering with measuring as real artificial intelligence resources tested.