Currents 083: Joscha Bach on Synthetic Intelligence

Jim talks with Joscha Bach about current and future developments in the generative AI space. They discuss the skepticism of the press, small productive applications, questions about intellectual property rights, confabulation in human thinking, nanny rails, 3 approaches to AI alignment, Aquinas’s 7 virtues, issues of consciousness-like agency, love as an answer to the alignment problem, the difficulty with fairness, serving shared sacredness, dealing with entropy, integrated information theory & its incompatibility with the Church-Turing thesis, neural Darwinism, a point where extrapolation & interpolation become the same, building an AI artist, free will, the capacity of human memory, consciousness as a conductor, the scaling hypothesis in AGI, making the system learn from its own thoughts, computation as a rewrite system, neurons as animals, and much more.

Joscha Bach is a cognitive scientist working for MIT Media Lab and the Harvard Program for Evolutionary Dynamics. He earned his Ph.D. in cognitive science from the University of Osnabrück, Germany, and has built computational models of motivated decision making, perception, categorization, and concept-formation. He is especially interested in the philosophy of AI and in the augmentation of the human mind.