Artificial intelligence and humans: basic models of relationships in science fiction

Abstract

Many of the problems raised by artificial intelligence (AI) researchers between 2000 and 2020 have, in one way or another, already been addressed in science fiction before. Analysis of science fiction texts allows us to identify the following models of relationships between humans and AI. The first model can be called exclusively friendly AI (R. Heinlein, “The Moon is a Harsh Mistress”). Such AI is a friend, an assistant to a person in all his affairs – including those directed against other people. The second model can be characterized as “friendly AI with built-in ethical restrictions” (A. Azimov, “I, Robot”). The third model is a neutral AI, weakly interested in human affairs, located “above good and evil” (S. Lem, “Golem XIV”). The fourth model is AI, which has its own goals that diverge from the goals of humanity and is therefore potentially hostile (A. Clark, “2001: A Space Odyssey”). The fifth model is an AI openly hostile to humanity (“Terminator”).