Table of Contents
In the midst of declarations of a new boom in artificial intelligence (AI), a roundtable discussion involving Nozomi Ohmori, book reviewer and science fiction (SF) translator; Satoshi Hase, SF writer; and Yoichiro Miyake, AI developer, was held at Genron Café. The theme of discussion was the current state of AI development and its relationship with SF’s imagination.
Miyake began the discussion by summarizing the history of AI. AI development focused on technologies that handled symbols like natural language processing until recently. However, today, technologies such as image recognition handle things other than just symbols.
Miyake also introduced the idea of real-world-oriented AI, which produces data related to the real world so that it is more recognizable to AI. In response to this, Hase submitted examples of smart cities developing in African urban centers and China, and pointed out that issues such as privacy and data governance could be raised by real-world-oriented AIs.
Miyake noted the interesting similarity between a type of game AI and smart cities. It is known as “meta-AI,” which controls all the characters’ behaviors in a given game.
In contrast, character AI, which controls individual characters, is made up of several modules and decision-making systems. Ohmori pointed out that characters in a novel resemble character AIs. Hase argued that events such as disasters in the climax of entertainment novels play a role similar to meta-AI, which coordinates the movements of character AIs.
Miyake admitted his study is highly influenced by SF’s imagination. This includes various other images of AI which are different from those developed in the mainstream of AI study in Europe or US, which only defines intelligence as the capacity of symbols and languages’ control. Ohmori pointed out that some important authors from Isaac Asimov to Ted Chiang came from the periphery of European culture and that their imagination could be influenced by their origins.