Book review: ‘Artificial Communication’ by Elena Esposito

When it comes to the algorithms that work with deep learning and big data there’s a strange paradox emerging, says Elena Esposito in ‘Artificial Communication’ (The MIT Press, £22.50, ISBN 9780262046664). The better they become at driving cars, composing music and scanning books, the more our discomfort increases. You only have to type an email or peck at a text to find that the untrustworthy predictive text of yesteryear has given way to a spookily accurate set of suggestions about what your next word might be in your linear narrative. Or even how to complete your sentence. This eerie feeling of machines or software behaving in a way that’s too similar to our own human thought processes has given rise to the expression ‘uncanny valley’.

Esposito, who is a professor of sociology working in the field of social systems theory, argues that when we think along these lines – ‘how does my smartphone know what my favourite songs are?’ – we are drawn into wondering if machines have simply become too intelligent. But in her latest book, subtitled ‘How algorithms produce social intelligence’, she argues that this sort of comparison is misleading.

To get a clearer picture of what’s going on, we need to turn the telescope around: if machines contribute to social intelligence, it will be not because they have learned to think like us, but because we have learned to communicate with them. The strength of this idea lies in the proposition that to understand our interaction with machines we need to stop thinking so much in terms of artificial intelligence by shifting the emphasis to artificial communication. After all, the possibility that we may not be talking to a human when we book flights, buy tickets and pay bills online has rapidly shifted into the area of probability, if not certainty.

The way we look at this interaction is flawed, says Esposito. When we interact with a ‘smart’ program we start to wonder if it might be intelligent, if only in a different way to humans. The wisdom of following this train of thought, she argues, is questionable: after all, communication is something that has always evolved, while what defines human intelligence remains something of a ‘mystery’. Meanwhile, the “information generated autonomously by algorithms is not random at all and is completely controlled, but not by the processes of the human mind”. The real challenge in machine learning and big data today, she says, is getting a steer on how to ‘control this control’, how to manage their impact in a global society in which uncertainly is for humans a way of life.

It’s a huge subject that Esposito examines by analysing the use of algorithms in different areas of social life, where machines are not the enemy pitting their superior wits against mere humans. It’s just that algorithms don’t think like us. Thought-provoking and profoundly relevant.