Prediction Machines, Updated and Expanded Hardcover – 25 December 2022

The book was delivered is awesome conditions.The book is a must read. You get up to date with everything going around us regarding AI and Deep Learning. It starts by giving a basic understanding, how we got here and what next could happen.The best part is that you can develop a sense of understanding of these things and predict in your own what might happen nextIf you're thinking of buying it, please do buy it

A very good view on AI from a economic lens

As a concept it's not bad, and has a few useful cases that a business enthusiast or data cruncher would love to read about. But the book mistakes current advances in ai to be about prediction by making a generalisation that sounds ok in English, but is wrong in ml. They define prediction as an ability to estimate what's missing. But completing a sentence with missing words is not the same thing as predicting the temperature tomorrow given the last few years, although both things are *missing. * technically, one is very different from another, and in many cases, the latter is almost impossible to predict. Eg: when you may be likely to fall ill next, given your vitals and past illnesses. The book misses this key distinction in prediction and assumes all kinds of prediction will become excellent and cheaper. Hence it makes a few erroneous statements, and repetitive chapters themed around those. However, it does well to separate prediction from judgment in all kinds of economic and social activity, and though neither of them are easy in most such activities, it stresses more on developing good judgement worth paying for. Cannot argue against that good advice, but can't recommend that you take the premise of prediction becoming ubiquitous and cheap seriously either. Next edition might benefit from more editorial scissors to trim out repetitions, and a deeper look at different types of predictions. Maybe recruit a scientist as a 4th author. Economists take too many things for granted :)