|
Imagine the following exchange taking place between a theatre goer (Alan) and the box office Agent (Ann):
To appreciate the challenges one must meet in answering these questions, imagine trying to get a computer to engage in a conversation like this. First, the computer must ignore signals like umm and er. If the computer thinks er stands for the word a, then it will have problems in making sense out of the rest of the sounds, so that they fit together to form a sentence. Second, the computer must fill in what was meant but not said. When Ann says The closest two seats are separated by three rows, she means I can sell you tickets that entitle you and one other person to sit in two seats at the opera this Saturday, and these seats are the best in the house, and they are separated by three rows, and I cannot sell you tickets in the highest price range that enable you and one other person to sit any closer together than this. Saying all of that explicitly sounds stupid because people know how to work out what other people mean. They're good at doing this, because they know so much about the world, and the situation surrounding the conversation they're engaged in. Computers, on the other hand, aren't good at working out what people mean, because they lack this knowledge, unless we tell them about it in explicit programming languages. Third, the computer must recognise why Alan said what he did. Responding with But we don't sell furniture, only tickets to the opera, or Oh really? I was wondering whether to go to Ibiza or Antigua for my holidays this summer would not be particularly helpful. Rather, Ann offered a piece of information that Alan didn't ask for: that he could buy two adjacent tickets at the rear of the stalls. At the moment at least, computers are unable to anticipate needs in this manner, and so human-computer communication is very different from human-human communication.
In light of these challenges, it has become evident in recent years that a range of disciplines contribute to a modern understanding of the factors which determine human communication. First, Artificial Intelligence is concerned with simulating intelligent behaviour. On the face of it, one might not think that engaging in the above conversation requires intelligence. However, when one looks deeper, it becomes clear it does. Both Alan and Ann must use what the other says to reason in quite complex ways about what the other person wants and believes. If they don't work out what's going on in the other person's head, they won't respond in the appropriate way. Second, Computer Science contributes to our understanding of the fundamental concepts of information processing, and in many cases human communication is a passing of information from one person to another. Third, Linguistics is concerned with modelling the rules and principles that underlie human language abilities. Fourth, Psychology uses methods of observation and controlled experimentation in analysing behaviour and experience. Fifth, Logic studies both artificial and natural languages by characterizing what consequences follow from the truth of assumptions; it provides a means of connecting language to the world. And finally, Philosophy provides various ways of thinking about the concepts of language, mind and communication themselves. Cognitive Science aims to integrate these diverse concepts and methods into computational analyses of human behaviour.
|
Informatics Forum, 10 Crichton Street, Edinburgh, EH8 9AB, Scotland, UK
Tel: +44 131 651 5661, Fax: +44 131 651 1426, E-mail: school-office@inf.ed.ac.uk Please contact our webadmin with any comments or corrections. Logging and Cookies Unless explicitly stated otherwise, all material is copyright © The University of Edinburgh |
