Oxford Lecture: What Will A Companionable Internet Agent Be Like?

About

The lecture begins by looking at the state of the art in modeling realistic conversation with computers over the last 40 years. The talk argues that there has been real progress, even though some systems of the late 1960s were remarkably good, a fact largely forgotten now.

I then move on to ask what we would want in a conversational agent that was designed for a long-term relationship with a user, rather than the carrying out of a single brief task, like buying a railway ticket. Such an agent I shall call ‘companionable’ and I shall distinguish several functions for such agents, but the feature they share will be that, in some definable sense, an artificial Companion should know a great deal about its owner – derived both from conversation and from the internet itself – and can use that information.

For this lecture, it will not be important what form, robotic or otherwise, a Companion has and I shall not focus on developments in speech understanding and generation but just assume the state of the art. The focus will be, first, on the technical issues of what such a Companion should know and how it can gain and use such knowledge though the understanding of conversations and searching the internet; and, secondly, on what the social implications of such Companions will be: will we trust them, will a Government or their manufacturer demand access to what they know about us, will they talk to each other about us, and what will happen to their unique knowledge of us when we die?