Cade Metz

AI is such a juggernaut today that it's hard to remember how little respect and attention it got in the 1980s and 1990s among computer scientists generally. I began my career in earnest then, and no one I knew in academia or industry was working in the field. After some signal failures to deliver in the 1970s, the entire field fell into disrepute.

Metz does an exceptional job of chronicling the research that changed all that, and especially the key people who stubbornly stayed focused on the work. He correctly highlights the key technical contributors as well -- advent of huge amounts of data, enormous distributed storage and compute capacity, the happy accident of GPUs designed for rendering video games working amazingly well on the math required by machine learning. It's all written in a really accessible way. He explains what convolutional neural networks are in a way that an ordinary person can understand.

The book discusses the tension between folks who believe in artificial general intelligence and those who think that accomplishment is in the distant future. The people debating that point, and doing the research, talked to Metz, and he uses their words directly to explain the different points of view.

This is an excellent history, taking the field right up to the present day. No doubt there will be plenty of fodder for a sequel, in ten or twenty years!… (mais)