Dissertation Summary

October 15, 2024

TL;DR

This is a summary of my dissertation. I wrote it in 2023. It was created by NotebookLM along with an NPR style podcast you can listen to.

Generative AI and its Implications for Policing

Author: Michael Zidar

Main Themes

Generative AI and Policing

The dissertation explores the potential impact of generative AI on policing practices, examining its potential benefits and challenges. It argues that this technology has the power to revolutionize information access, analysis, and decision-making within law enforcement agencies.

Historical Technology Adoption in Policing

The study analyzes three decades of data from the Law Enforcement Management and Administrative Survey (LEMAS) to understand how police departments in the US have adopted new technologies over time. This historical perspective is crucial for anticipating how generative AI might be received and implemented in the future.

Technology Adoption Maturity Index

The dissertation develops a Technology Adoption Index based on LEMAS data to quantify the level of technology adoption across police departments over time and geographical regions. This index reveals trends and disparities in technology adoption, informing future strategies for AI implementation.

Agent-Augmented Practice (AAP) Framework

Recognizing the transformative potential of AI, the dissertation proposes an Agent-Augmented Practice (AAP) framework to guide the effective adoption and integration of generative AI into policing workflows. This framework emphasizes a collaborative approach, where AI agents augment human capabilities rather than replacing them.

Key Ideas and Facts

LEMAS Data Analysis

  • The study analyzes data from 1987 to 2020
  • Encompasses responses from approximately 3,000 policing agencies across the US

Analysis of LEMAS data reveals an upward trend in technology adoption by policing agencies since the 1990s. The study identifies various technology-related themes present in the surveys over the years, including:

  • Digital records
  • Communication systems
  • Crime analysis & mapping
  • Novel technologies like gunshot detection systems

Geographical Disparities

The Technology Adoption Index reveals disparities in technology adoption across US states and regions. For example, in 2020:

  • Florida had the highest average normalized tech adoption score
  • South Dakota had the lowest

Generative AI's Potential

The dissertation argues that generative AI can significantly enhance police work by:

  • Automating report writing and data entry tasks
  • Improving access to and analysis of policy documents and legal information
  • Facilitating real-time data analysis and visualization
  • Providing officers with contextually relevant information during field operations

Agent-Augmented Practice (AAP)

The proposed AAP framework emphasizes a collaborative relationship between human officers and AI agents. This approach aims to enhance police practices through AI assistance while retaining human oversight and ethical considerations.

Important Quotes

"This dissertation is an attempt to understand the readiness of police departments to adopt the generative artificial intelligence transformation currently occurring."

"To anticipate the future implications of generative AI in policing, a deep understanding of past trends is necessary."

"This dissertation posits that the AI-empowered capacity to recall, summarize, and interpret information will fundamentally reshape how police officers engage with their communities."

"The principal argument herein is that the IT practitioner and the IT researcher are distinct entities, akin to the police officer and criminal justice researcher."

Recommendations and Future Work

  1. Develop consistent technology-related questions for future LEMAS surveys to facilitate longitudinal analysis
  2. Conduct deeper investigations into how technology is used in core police functions
  3. Refine the Technology Adoption Index using model-based approaches for more accurate measurement
  4. Research and develop AAP-based solutions tailored to specific policing needs and contexts

Limitations

  • LEMAS data, while extensive, might not capture all nuances of technology adoption in policing
  • The study primarily focuses on quantitative analysis of technology adoption, with limited qualitative insights into the experiences and perspectives of police officers
  • The AAP framework, while promising, requires further development and empirical testing before implementation

Overall, this dissertation provides a valuable contribution to understanding the historical context and future potential of AI in policing. It underscores the importance of responsible AI adoption guided by a clear understanding of both the opportunities and challenges this technology presents for law enforcement agencies.