The LCO Issue Paper, AI Case Study: Probabilistic Genotyping DNA Tools Used in Canadian Courts considers the role and impact of AI-driven probabilistic genotyping (PG) technology to generate evidence used in criminal proceedings in Canada.
PG technology uses artificial intelligence algorithms to analyze DNA samples collected in police investigations or criminal prosecutions. This paper, authored by criminal defence counsel Jill R. Presser and Kate Robertson, examines whether and how AI-driven technologies like PG can continue to meet the high standards of due process, accountability and transparency, and fundamental legal rights demanded by the Charter and, by extension, the criminal justice system.
Like many AI and algorithmic tools used in the justice system, PG tools have human rights, equity, due process and access to justice consequences. Absent proper scientific study, regulation, and enforcement of relevant Charter rights and due process protections, PG DNA evidence may lead to wrongful convictions. There is also a risk that AI tools, including PG DNA algorithms, will worsen racism in Canada’s justice system and put access to justice further out of reach for many Ontarians. For these reasons, this report makes several recommendations, including
- Statutory amendments focused on the use of PG DNA analysis as evidence.
- Statutory amendments focused on enhancing systemic transparency and accountability.
- Improved practices and training.
- Review of legal aid programs to identify and remedy gaps in policies and budgetary constraints.
- Further Research and Evaluation.
An Executive Summary of the report is available.
- AI Case Study: Probabilistic Genotyping DNA Tools Used in Canadian Courts (June 2021)
- Executive Summary, AI Case Study: Probabilistic Genotyping DNA Tools Used in Canadian Courts (June 2021)
In March 2022 the LCO was pleased to receive a detailed response to this paper from the Canadian Scientific Working Group on DNA Analysis Methods (also available in French).
The Working Group is a platform for discussion and sharing of Canadian best practices in forensic biology with representation from four entities: the Centre of Forensic Sciences in Ontario, the Laboratoire de Sciences Judiciaires et de Médecine Légale in Quebec, and the National Forensic Laboratory Services and National DNA Data Bank of the RCMP.
The letter provides additional information on the workings of probabilistic genotyping technology; how it is validated, audited, and the kinds of information about its workings which may be disclosed and tested in a court proceeding; the process of interpreting results against a hypothesis of the case; and suggests distinctions between “mathematical” and “artificial intelligence” systems. The letter concludes by acknowledging the challenge of presenting scientific evidence to courts and the important role of expert witnesses in explaining it.