Work

The Role of Artificial Intelligence in Criminal Investigations: Challenges of Bias, Transparency, and Legal Accountability

Year: 2023
Type: article
Abstract: From facial recognition to predictive policing, automated analysis of digital evidence and probabilistic forensic software to risk scoring, AI has become part of the day-to-day of criminal investigation. They can deliver speed and uniformity, but they're also been responsible for misidentifying suspects, issuing incorrect sentences, and controversy over who's to blame when an algorithm is used to produce an unfair result. This article focuses on three related issues – bias, transparency, and legal accountability. It examines empirical studies, doctrinal writing, and regulatory documents from the United States, the United Kingdom, and the European Union and finds that these issues are not isolated flaws, each of which warrants a remedy in isolation. Bias creeps in at each stage of the investigative process, opacity makes it hard for defendants and courts to identify it, and current legal doctrine assigns blame at the right moments where there are holes in the investigative process. The article proposes a 3-phase accountability model (ex ante authorisation and validation, in operation oversight, ex post disclosure and remedy) and specifies the safeguards needed for each phase. It holds that the legality of the use of AI in investigations is not tied to the accuracy of any one intelligence system, but to maintaining the capacity of people to challenge the product of the intelligence systems.
Language: English
Subfield: Law
Open Access status: diamond
OpenAlex ID: W7224054033
FWCI:
Cites: 21
Cited by:
Related to: 10