Book review number 250

Goodreads book review

This is my two hundred fiftieth review on Goodreads since 2010. You can read them all, fiction, non-fiction, and all sorts of other stuff, at this link:
Rich Magahiz’s reviews on Goodreads

You Look Like a Thing and I Love You: How Artificial Intelligence Works and Why It's Making the World a Weirder Place

You Look Like a Thing and I Love You: How Artificial Intelligence Works and Why It’s Making the World a Weirder Place by Janelle Shane
My rating: 4 of 5 stars

This was a solid piece of work on the state of machine learning which is so rapidly changing that one of the main platforms she uses in her examples, GPT-2, has already been replaced by another one with capabilities a hundred or more times that of its predecessor. So while it gives a good account of the challenges in applying the technology, it will only be a snapshot in time of how far researchers have been able to go (until it is revised perhaps some day). It draws on her popular blog with examples of her own experiments along with descriptions of what other researchers have been working on as broadly as can be accommodated in a work aimed at a general audience. Although I do have some involvement with deep learning in my job, my own interest is that of someone who is interested in the technology in a general way.

The book does a pretty good job at depicting the attraction of machine learning AI systems and the well-known pitfalls along the way. There is a discussion about the way bias can creep in even though it is not part of the model design itself, and thoughts about what areas are suitable for applying the technology and what ones are still a long way off from practicality. Throughout there is a big heap of the trademark humor and irreverence readers of the blog and of the Twitter stream know about along with the little illustrations by the author which liven even the heaviest topics up. I did not get a sense that the book attempts to be a cheerleader for AI or portrays a rosy future for how it will change our lives, instead modestly stressing all along that the observations are based on neural networks comparable in complexity to that of a worm, rather than that of say of some kind of insect as of the current time of this review.

comments powered by Disqus