My work in AI started in 1979. After receiving my PhD from Johns Hopkins University, I moved to Connecticut to do postdoctoral research with Professor Roger Schank in the Department of Computer Science at Yale University. At the time, Yale had a leadership position in the burgeoning AI subdiscipline of natural language processing.1 It was a beehive of activity. Each of Roger’s many graduate students was attempting to build computer systems to perform tasks such as machine translation (i.e., translate text from one human language into another human language), question answering, and summarizing news stories. Each week, we had a well-known academic come to town and present to the group. Roger put me in charge of taking the speakers to dinner, so I got to hobnob with many academic celebrities. It was a wonderful two years for me.
In August 1981, I accepted a teaching position at Brandeis University and was getting ready to leave Yale. At the same time, Roger had just received funding to start an AI company named Cognitive Systems. He approached me to join him in this new venture and told me that, even though I would likely be a successful academic, based on my personality, I would not enjoy it and that I would be much happier in the commercial world. I saw his point, so, with regret, I backed out of the Brandeis position at the last minute (they were understandably unhappy) and joined Roger at Cognitive Systems. As things turned out, it was the best career advice I have ever received.
In the early 1990s, I created Esperant, a natural language system that became a leading business intelligence product. More recently, I cofounded Device42, a company that is emerging as a market leader in IT infrastructure analytics. I am also a successful angel investor with a portfolio that includes many AI companies and one unicorn.
Through all these years, I’ve grown frustrated at the fear-inducing hype around AI in popular culture and media and at the overstatement of AI’s capabilities from its vendors. It’s fair to say I have a good understanding of AI, how it works, and what it can do. My goal in this book is to provide you with some of that knowledge. I will keep the technical detail to a minimum, and we’ll discuss whether any of that fear is justified.
Figure 0.1 The different types of AI.
The image in figure 0.1 is a diagram of the different types of AI. By the end of the book, you should have a high-level understanding of how each of these types of AI work. You will also learn why none of these types of AI will progress into the types of AI that people fear.
If you are among the more technically inclined, you may prefer a more in-depth treatment of some of the topics in this book. If so, you can find it on my website, www.aiperspectives.com. The site provides hundreds of pages of technical detail in a dense, textbook-like format. It is not as easy to understand as this book, but I have worked hard to make it more accessible than many AI textbooks by leaving out the advanced mathematics found in them.