Data Governance for Justice and Human Rights: Forensics, Flow, and Frontiers

This chapter introduces the volume as a whole, outlining the current and future directions in the legal use of data, and exploring how data can be utilized in the legal system to promote justice and human rights, as well as enhance the legal decision-making process itself.

Part I examines some of the ways in which Artificial Intelligence (AI) and Machine Learning (ML) technologies are marshalling data to improve legal decision-making. The included chapters examine the ways in which data science has been used to improve legal decision-making, from its inception in the mid-twentieth century to Groningen’s school's more recent work on using AI to model the fact-finding process itself. Data, when governed judiciously, can be used to improve legal decision-making and thus improve access to justice, better justify decisions, reduce biases in decisionmaking, and evaluate evidence to improve accuracy and reduce miscarriages of justice.

Part II examines new strategies to govern the use of data in the legal system: identifying and addressing cognitive biases among forensic scientists, regulating the forensic use of bioinformatics in the criminal justice system, and eliminating the use of false legal information hallucinated by AI systems.

Part III of this book describes some novel forensic applications of genetic data, particularly in the field of bioinformatics and its advances in human identification. Some of the most significant advances have been made in the field of bioinformatics and its application to human identification. The chapters in this section focus on these developments, examining how advances in SNP sequencing, combined with computational methods for kinship identification, are leading to the clearance of previously unsolvable cases.

This chapter reviews the use of data science and machine learning applications to facilitate the analysis of evidence and the processes of juridical proof. Innovations in data science and machine learning have led to advancements in forensic interpretation, risk prediction, and the efficient processing of large datasets of ‘open source’ material. However, the ‘holy grail’ of data science development in the evidence domain is the modelling of legal arguments themselves. Data science holds the potential to play a leading role in understanding and modelling the fact-finding process. Argumentation schemes have thus become a major research focus and hold the potential - when aligned with parallel developments in legal reasoning – to fully automate the trial process. This chapter considers the standard applications of data science to legal evidence – forensics, prediction, and data processing - before reviewing the work of the Groningen School, which has played an important role in shaping this field. Crucially, it is intended that this introductory chapter offer an integrated perspective on evidential reasoning, demonstrating not only the unique affordances of particular inferential modes encountered in the surveyed literature but also highlighting the compound and contextually dependent nature of these features. This chapter, therefore, highlights the importance of attention to conditioning factors based upon the structure of, and inter-relations between, the overall form of the legal system, legal sub-field, mode of argumentation, and form of adjudication.

This chapter discusses the application of artificially intelligent systems to the assessment of evidence in criminal cases and proposes a system for formalizing argumentation schemes for machine learning. It draws on recent work on argumentation schemes and adapts them to address common issues encountered in criminal cases, such as the credibility of witness testimony, the probative value of similar facts, and the evaluation of expert evidence. This is illustrated by systematically setting out the arguments presented over the course of several decades in the trials of Robert Earl Hayes and identifying critical questions to evaluate the evidence. Hayes’ case is a well-known miscarriage of justice that has been extensively investigated and litigated. It therefore serves as a good model for evaluating the arguments put forward in the case over the course of several decades, as the facts of the case are well-known and reliable. The proposed system combines argumentation schemes with scenariobased reasoning. The arguments are separated into different scenarios put forward by the parties. Each scenario is evaluated separately; in the final step, their relative plausibilities are compared using an abductive argumentation scheme to evaluate the most plausible scenario. The final result is a qualitative plausibility value that signifies how justified we are in believing that a given scenario is the best explanation for the available evidence, at the time of evaluation. The plausibilities proposed here are qualitative, relative and are correlated with legal standards of proof. This system assists in identifying and removing prejudices and cognitive biases that impede reliable evaluations of evidence. It also assists in better formalizing and defining argumentation schemes to evaluate evidence in criminal cases, so as to prepare the way for future computational applications. 

AI hallucinations pose an existential risk to existing laws and social order. The authenticity of documents in general and legal documents (e.g., statutes, regulations, bills, caselaw) in particular, is seriously jeopardized if Large Language Models (LLMs) produce – and we accept – false legal precedents. With the rise of the “AI empire”, humans must secure our societies’ legal foundations. This chapter will explore the risks of AI hallucinations and will provide a solution based on a document authenticity verification system. This is crucial to our survival as a society and should be built into all AI processors that deal with legal documents. Governments should enact legislation that requires AI systems to include methods of verification. This is fundamental to the rule of law in the age of AI.

The research concerning the effect of cognitive bias in forensic science decision-making has been well documented. Numerous studies across legal and forensic investigations have shown a number of distinct ways that expectations, motivations, and beliefs can influence memory, as well as the selection and evaluation of evidence. In human identifications, studies have shown how contextual information can affect the interpretation of skeletal remains when establishing a biological profile. Although the concept of cognitive bias is now well established, there is arguably still a lack of empirical research looking at possible mitigation strategies across different fields of forensic science, including the human identification field. This chapter discusses some of the empirical research conducted on cognitive bias and forensic decision making, with a focus on human identification. Mitigation strategies and proposed solutions are explored and discussed, along with future avenues.

Investigative Genetic Genealogy (IGG) has emerged as a premier investigative tool for identifying human remains and perpetrators of violent crimes where other methods have failed. As the number of cases resolved with the help of IGG has increased, there have been concomitant moves to regulate the field by applying existing regulations where possible and creating new ones to address issues specific to IGG. Regulation has come from all levels—from the Department of Justice, state legislators, judges, district attorneys, non-profit organizations, forensic laboratories, and IGG practitioners themselves—and consists of direct and indirect regulation, and binding and non-binding rules. This chapter provides an overview of all forms of IGG regulation to date and traces a few possible directions for the future of IGG regulation.

Investigative Genetic Genealogy (IGG) is a powerful technique that combines DNA analysis with traditional genealogical research to narrow down the possibilities for the identity of an unknown individual. Since the arrest of the Golden State Killer in May 2018, IGG has been publicly acknowledged in 464 Unidentified Human Remains (UHR) and 318 perpetrator cases as of December 2023, and has been used in many more unacknowledged cases. Parabon alone has contributed to over 320 identifications, more than 1/3 of which have not been acknowledged publicly. A critical component of successful IGG is the quality of the genotypes generated from the forensic sample. This chapter will provide a brief overview of the IGG process, describe how data quality impacts IGG cases, and discuss various laboratory and bioinformatics techniques for generating high-quality genome-wide Single-Nucleotide Polymorphism (SNP) genotype data from challenging forensic DNA samples.

Identification of unknown persons, be they missing or unidentified, is a significant challenge, an important humanitarian responsibility, and a fundamental part of the healing process for families and communities. The only resolution for the families (and friends) who have lost a relative is credible confirmation of the death of their family member (and friend) and that the remains of their relative (and friend) will be treated with dignity, in accordance with their culture, religious beliefs, and the laws of their respective countries. Forensic evidence can play a significant role in the humanitarian and investigative processes. Of the various forensic tools to assist in the identification process, DNA is particularly versatile because it can be found in all tissues of the human body and thus does not rely on a specific target tissue, as do, for example, fingerprints and dental records. Moreover, in the absence of antemortem samples or records, DNA identification can be carried out by kinship analyses as long as family reference samples are available. The concepts and foundations of relationship testing are well-established and described herein. Current DNA typing capabilities have been a mainstay in unknown person identifications and in disaster victim identifications. However, DNA technologies and databases have advanced substantially in the past few years to enhance the scope and reach of kinship analyses. Expanded technology capabilities (high throughput sequencing and concomitant bioinformatics) and additional types of DNA databases (genetic genealogy) enable higher accuracies in kinship testing, association with more distant relatives, and greater typing success due to analysis of lower quantity and quality samples. Combining kinship data with access to public records and enhanced bioinformatic and computational power enables forensic investigative genetic genealogy to provide better success of humanitarian services for families and communities, as well as identification of the source(s) of evidence found at crime scenes. 

This chapter reviews developments in forensic DNA that enable case clearance using genetic data from rootless hair. Advances in Massively Parallel Sequencing (MPS) have enabled the production of DNA profiles of Single-Nucleotide Polymorphisms (SNPs). The resulting SNP profiles can then be used to identify persons of interest – usually unidentified decedents or suspects in criminal cases – using Forensic Investigative Genetic Genealogy (FIGG). FIGG uses genetic genealogy databases of SNP profiles to generate relatedness estimates to identify persons of interest. Once candidates are identified through FIGG, the identification is normally confirmed through the direct comparison of Short Tandem Repeat (STR) forensic DNA profiles – either through direct comparison for crime scene samples, or through kinship testing for human remains. However, SNP profiles can be generated in cases in which the DNA evidence exists in too low quantities or is too degraded to generate a traditional (STR) forensic DNA profile. This allows for cases to be cleared, even where an STR DNA profile is not available to be uploaded and searched against CODIS databases. Novel statistical methods are available for direct comparison between SNP profiles, which enables the probative value of the evidence to be assessed by triers of fact in criminal cases. In this chapter, we focus on the clearance of cases using rootless hair, as these samples are commonly found in criminal and human remains investigations but have resisted conventional DNA analyses. Forensic DNA analysis is therefore poised to move beyond CODIS, and this poses novel challenges for courts and legal professionals.