Title: Living with Intelligent Machines

Speaker: Nello Cristianini (University of Bristol)

Location: Building D6 Room 1.18 – Via delle pandette, 9

ABSTRACT

The way we build intelligent machines today involves two components: a (fairly) general learning algorithm, and a (very) large set of training examples. These two elements are combined to create machine translation, computer vision, online recommendations, spelling correction, and so on. In many cases we know that the intended behaviour can be closely emulated, but we have done away with modeling the mechanism that generates it, replacing that with statistical correlations. As the training examples are often obtained “from the wild”, they might contain information that we are not aware about, including various cultural biases. If these biases enter in the statistical models, they can be difficult to detect, and become part of shaping the behaviour of the intelligent agent. As we train machines by exposing them to media content and other samples of human behaviour, and these machines have the capability to emulate those behaviours, it is important that we understand the biases present in the data: this is one case in which computational social sciences become an important element of AI design. Other cases include the automation of psychometrics, and the risk of addiction. The social, ethical and legal consequences of using modern-type of AI can be better understood and managed when we think about AI also from a social science perspective. 

BIO

Nello Cristianini is Professor of Artificial Intelligence (AI) at the University of Bristol. His current research covers the large-scale analysis of media content (news and social media), using various AI methods, the design of new AI methods, their application to digital humanities and computational social science, and the social impact of Big Data and AI technologies. Cristianini is the co-author of two widely known books in machine learning, An Introduction to Support Vector Machines and Kernel Methods for Pattern Analysis, as well as a book in bioinformatics, Introduction to Computational Genomics. He is also a recipient of the Royal Society Wolfson Research Merit Award and a current holder of a European Research Council Advanced Grant. In 2014, Thomson-Reuters included him in a list of the most influential computer scientists of the decade. Before joining the University of Bristol, he has been a professor of statistics at the University of California, Davis. Currently he is working on social and ethical implications of AI.