The Data Science Handbook, 2nd Edition

6Overview: Machine Learning and Artificial Intelligence

Machine learning and artificial intelligence (ML and AI) obviously play a large role in data science, and increasingly in society in general. But, they are large topics with a lot of moving parts, and it’s easy to get lost in the weeds. This brief chapter zooms out and gives a high‐level, conceptual overview. It discusses the relationship between ML and AI, their relationships with other fields like statistics, the overarching paradigms that pervade ML and AI, and how the professional landscape is evolving in response to the latest technical developments. In subsequent chapters, I will get into the technical details; the goal here is to give you the big picture that will frame future discussions.

6.1 Historical Context

“Machine learning” was partly born out of the initial failures of the AI movement in the 1960s, 1970s, and 1980s. For a long time, people were very focused on the idea that computers could be made to think, and it was widely expected that thinking machines were only a few years away. There is an anecdote that Marvin Minsky, one of the founders of AI, once assigned a grad student the task of working out computer vision over the course of a summer. People were thinking about the human brain as a big logic engine, and a lot of the focus was on getting computers to mimic the logical processing that humans do.

AI failed (at least relative to the hype it had generated), giving rise to an “AI winter” in which ...