One idea the science of intelligent systems needs

The dual quests ??" to understand intelligence as it arises in biology and to build it in machines of our own making ??" have been with us for nearly a century. While neuroscience and psychology seek to characterize intelligence in terms of underlying neural mechanisms and emergent behavior, we still lack a set of principles that can account for how the interactions of many neurons within a brain gives rise to intelligent behavior. Meanwhile, modern AI is following a largely empirical path to create a new technology without scientific foundation. While the resulting algorithms exhibit intriguing behavior, the working principles are unknown, usage risks are incalculable, and the path to further development is unclear. The premise of this seminar is that the current century will see the emergence of a new scientific discipline ??" the Science of Intelligence ??" that seeks to elucidate universal principles that govern intelligent systems, be they biological or machine, and that this will provide both much needed insight in neuroscience and a solid foundation for technological development. The seminar will meet weekly, featuring speakers from across campus and elsewhere, who will share their views on the open questions, topics of study and debate, and relevant disciplines that a science of intelligence will encompass.

Speaker: Steven Piantadosi, UC Berkeley