Building Criminological Theory Using Artificial Intelligence?

• In Event: Rutgers at 50: Theoretical Contributions to Criminology

Abstract

Artificial intelligence tools have the potential to advance the general field of criminology and criminological theory, in particular. However, to explore this avenue of scholarship it is crucial to unpack what AI actually is and how it works. Tools such as Apple’s Siri or Amazon’s Alexa for example rely on training to retrieve information based on user input, but they do not necessarily create information. Generative AI programs such as ChatGPT or DALL-E are the kind that “learn” and can create original responses without further intervention from humans. Machine learning programs have proven useful for developing hypotheses and solving problems in several disciplines including quantum physics, chemistry, and medicine (Ananthaswamy, 2021; Blades, 2021). In this presentation, I explore the potential role of generative AI in the process of theory development in criminology. More specifically, I discuss the benefits and pitfalls of relying on approaches such as machine learning to not only test standard hypotheses, but also and perhaps more critically, to uncover new patterns of relationships and associations: At what point is AI “building” theory? What kind of “theory”? I draw on specific examples to explore these questions in greater detail and relate them to more general questions about the intersection of theory and technology derived from contributions by faculty and students at the Rutgers School of Criminal Justice and beyond.

Author

• Shannon Smith, Rutgers University