Can science fiction be written by or with generative artificial intelligence? Science fiction was deliberately chosen as the genre rather than crime fiction, romance, or others. Science fiction can deviate from descriptions of the real world in its portrayals, meaning AI cannot fail at describing reality. The race to the Moon was chosen as the theme. Abundant information is available on this topic that the AI was trained on, giving it the relevant background knowledge. In practice, the project resulted in 31/2 stories. The grandiose and widely held views of ingenious AI were put to the test. The paper explains working with AI and its specific characteristics. The linguistic style of AI responses is noteworthy. An emotional and seemingly understanding mode of expression suggests an intelligence that is not actually present. Current generative AI is a text generator based on probabilistic connections between words and word groups. The results are data constructs from word clouds that the user interprets. The meaning that the user perceives is not inherently contained in the AI's text output. Software companies have designed AI outputs to make humans suspect intelligence behind them. This sells better than raw data. The user is led to anthropomorphize so-called artificial intelligence. Extensive education about this intelligence effect is necessary. This book aims to make a small contribution to that end. It is not a scientific study, but rather an application of AI through examples in the nature of a case study. A novel about a novel: 1/2 Story: Draft for a science fiction about a fake Moon landing 1st Story: The Last Gunslinger - with an unusual ending 2nd Story: The Gunslinger is beamed into the future - a familiar concept 3rd Story: No fake Moon landing, just a small step - a plot ready for film It becomes clear how word clouds can lead to nested plots. However, in the third story, the AI loses track of the many deceptions it builds into the narrative. Other highlights

Spécifications