Is artificial intelligence a hazardous technology? Economic trade-off model

Abstract Artificial intelligence (AI) demonstrates various opportunities and risks. Our study explores the trade-off of AI technology, including existential risks. We develop a theory and a Bayesian simulation model in order to explore what is at stake. The study reveals four tangible outcomes: (i) regulating existential risks has a boundary solution of either prohibiting the technology or allowing a laissez-faire regulation. (ii) the degree of ‘normal’ risks follows a trade-off and is dependent on AI-intensity. (iii) we estimate the probability of ‘normal’ risks to be between 0.002% to 0.006% over a century. (iv) regulating AI requires a balanced and international approach due to the dynamic risks and its global nature.