In the fast-growing and ever-evolving world of artificial intelligence, it can take some real brains to distinguish realistic and potentially useful applications of AI from exaggerated claims about AI capabilities that do not work as promised or do more harm than good.
In Princeton, those brains are Princeton University computer science professor Arvind Narayanan and graduate student Sayash Kapoor, who together are the authors of “AI Snake Oil: What Artificial Intelligence Can Do, What It Can’t, and How to Tell the Difference,” published in September by Princeton University Press.
The two scholars, both named to TIME magazine’s inaugural list of the “100 most influential people in AI,” speak on their book and current topics in AI at the Princeton Tech Meetup’s gathering at Princeton Public Library, 65 Witherspoon Street, on Thursday, October 24, from 6:30 to 8 p.m. The event is free to attend. Register at www.meetup.com/princeton-tech/events/302682047.
The authors define AI snake oil and the problems it causes in the introduction to their book:
“AI snake oil is AI that does not and cannot work as advertised. Since AI refers to a vast array of technologies and applications, most people cannot yet fluently distinguish which types of AI are actually capable of functioning as promised and which types are simply snake oil. This is a major societal problem: we need to be able to separate the wheat from the chaff if we are to make full use of what AI has to offer while protecting ourselves from its possible harms, which in many cases are already occurring.”
The authors also note what they see as the root cause of pervasive AI hype and falsehoods: “Since we started working together,” they write in their introduction, “we’ve come to better appreciate why there is so much misinformation, misunderstanding, and mythology about AI. In short, we realized that the problem is so persistent because researchers, companies, and the media all contribute to it.”
Case in point: the series of events that inspired the book project in the first place, starting in 2019 when Narayanan was approached by a researcher at a hiring automation firm. The company was hiding evidence that its tool was not nearly as effective as advertised. This interaction became part of a lecture he gave, the slides from which he later uploaded online — where they went viral, underscoring the widespread interest in this topic.
To learn enough about the topic to feel confident writing a book on it, Narayanan first taught a university course on the “Limits of Prediction” using AI. Kapoor was a newly arrived graduate student in the course. Research conducted by both over the subsequent years became the basis of the book.
In addition to serving as a professor of computer science, Narayanan is also the director of the university’s Center for Information Technology Policy. His previous books include serving as co-author of “Bitcoin and Cryptocurrency Technologies” and “Fairness and Machine Learning.” He did his undergraduate work in India and earned his PhD from the University of Texas at Austin in 2009.
Prior to undertaking his PhD in computer science at Princeton Kapoor worked as a software engineer at Facebook, where he helped create AI for content moderation. He earned his bachelor of technology degree in computer science at the Indian Institute of Technology Kanpur.
“AI Snake Oil: What Artificial Intelligence Can Do, What It Can’t, and How to Tell the Difference” is available on Amazon.com. List price: $24.95.
