Machines will think: structure and interpretation of Alan Turing\'s imitation game

• Ano de defesa: 2021
• Autor(a) principal: Gonçalves, Bernardo
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Alan Turing; Artificial intelligence; Can machines think?; Experimento mental; Inteligência artificial; Jogo da imitação; Podem as máquinas pensar?; The imitation game; Thought experiment
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/8/8133/tde-10062021-173217/

Can machines think? I present a study of Alan Turing\'s iconic imitation game or test and its central question. Seventy years of commentary has been produced about Turing\'s 1950 proposal. The now legendary \"Turing test\" has grown a life of its own in the tradition of analytic philosophy with at best loose ties to the historical imitation tests (1948-1952) posed by Turing. I shall examine the historical and epistemological roots of Turing\'s various versions of imitation game or test and make the case that they came out from within a dialogue, in fact a scientific controversy, most notably with physicist and computer pioneer Douglas Hartree, chemist and philosopher Michael Polanyi, and neurosurgeon Geoffrey Jefferson. Placing Turing\'s views in their historical, social and cultural context, I shall reclaim their scientific and philosophical value for the sake of the discussion in the years to come. My study is organized according to three main philosophical problems whose analyses are backed by a subsidiary chronology of the concept of machine intelligence in Turing\'s thought (1936-1952). The first problem I will address is the identification of Turing\'s specific ambition which led him to announce that machines will think. War hero and brilliant mathematician, he challenged the conventional wisdom of what machines really were or could be and prophesized a future pervaded by intelligent machines which may be seen as a dystopia just as much as a utopia. I shall examine Turing\'s profile and take special interest in the way he was seen by his contenders. In the second problem, over and above the mere proposal of a test for machine intelligence, I will study Turing\'s proposition \"machines can think\" and its implied existential hypothesis -- \"there exists (will exist) a thinking machine\" -- from a point of view of the history of the philosophy of science. Unlike traditional readings of Turing, I found that Turing held a non-obvious realist attitude towards the existence of a mechanical mindbrain which he conjectured to frame the human and whose digital replica he intended to build in the machine. Turing\'s 1950 paper has been acknowledged as a complex and multi-layered text. Opposing views can be identified in the literature relative to the question on whether or not Turing proposed his imitation test as an experiment to decide for machine intelligence. I shall call this the Turing test dilemma and address it as my third and main problem. My findings suggest that Turing cannot have proposed his imitation game as something other than a thought experiment. And yet its critical and heuristic functions within the mind-machine controversy are striking.