Algorithms and humans

If I were recommending one book for a general reader about the brave new world of algorithmic decision making and AI in particular, it would be Hannah Fry’s Hello World: How to be Human in the Age of the Machine. The ground it covers will be familiar to AI/ML experts, but for anyone else this is a terrific and – importantly – balanced introduction to the issues. There are other books around that are great on the concerns – Cathy O’Neill’s Weapons of Math Destruction leaps to mind as one aimed at the general reader. But these tend to skate over the other side of the story. There are also somewhat denser reads, such as The Master Algorithm. But Hello World is a really clear, informative introduction.

Sandwiched in between a general intro and conclusion, the bulk of the book is a series of chapters covering different issues (the locus of decision-making power,  and the role of data) and domains (the justice system, medicine, cars, crime and art). There were points in most of the chapters that made me stop and ponder.

If you wanted to criticise the book, you’d probably say it’s great on the problems and questions, light on the solutions, but that would be most unfair. Algorithmic decision making sheds an unforgivingly bright light on the trade-offs and choices our society already makes. They look far more unpalatable when we’re forced to confront them in such a stark way – because the decisions are so fast, because they are not fudged, because machines are not inconsistent in their judgements, and so on. We are all grappling with the challenges. So if you have an uneasy feeling you ought to know more about this, Hello World is a good starting point.

[amazon_link asins=’0857525247′ template=’ProductAd’ store=’enlighteconom-21′ marketplace=’UK’ link_id=’0f1330b0-dba6-11e8-86b7-33d4704f88bc’]

One thought on “Algorithms and humans

  1. Think of it in terms of geography. Most of the human race can manage the hills and valleys of many places. But as the mountains get higher and further then fewer and fewer can do it. Question, if AI etc. are the Himalayas of decisions and their making, then how many have managed Mount Everest?

Comments are closed.