Tech Tent: In-depth on AI

In 2017 AI was everywhere, so in the last edition of Tech Tent for this year we set off to explore artificial intelligence. What is it, where is it making rapid progress, and what are the dangers for society if we get it wrong? For all the excitement about the supposedly rapid advance in creating intelligent machines, there are now plenty of dissident voices arguing that AI is overhyped. "The most rapid'progress' in so-called AI has been by downgrading what is considered AI so that any old algorithm or machine learning is said to be "intelligent". Our two special guests think wonderful things are happening in this field but agree that there is a problem in defining AI. "Everyone has a different definition and it's been constantly changing since the 1960s," says Tabitha Goldstaub, co-founder of the AI consultancy Cognitio She says computers have made great progress in some aspects of human intelligence, learning by experience in the same way a child does, to excel in tasks such as being able to distinguish between different human faces. But they then struggle with more arbitrary questions - AI is very good at doing a defined task better than a human but still isn't able to understand the logic that a baby can."