Response to Collins, Artificial Experts

Abstract

Harry Collins's basic idea is that machines can behave intelligently in domains where we have chosen to discipline ourselves into behaving like machines.' We use calculators and can work on assembly lines in a rigid, repeatable way because we have learned to respond in a way that is insensitive to context.2 Such behaviour is digitalizable. There will still be occasional breakdowns but we ignore them or unreflectively repair them. No type of computer, Collins contends, could behave intelligently in a domain where we act in a flexible, situationsensitive way. We can extend the domain of machine intelligence only be extending the range of machine-like human behaviour. And we can extend our robot-like activity without limits, although we may not care to do so.