As our justice system navigates advancing technology, we face a critical question: How should we incorporate facial recognition technology into our legal proceedings? Minnesota courts are currently contending with the admissibility of facial recognition as evidence. It is time to consider its role, not as a smoking gun, but as a valuable investigative tool.
What I observed as an expert witness at the Frye-Mack hearing in Ramsey County District Court reinforced this perspective. The evidence clearly showed that facial recognition technology has matured beyond its experimental phase. It's now a well-established field with robust scientific foundations, earning its place as a legitimate investigative tool.
The technology's current exclusion from courtrooms stems from concerns about its accuracy. Critics point to potential errors, particularly false positives where innocent individuals are incorrectly matched with criminal suspects. However, this perspective overlooks a crucial distinction: the difference between using facial recognition for absolute identification versus employing it as an investigative aid.
Consider how we already use other investigative tools. DNA analysis and fingerprint matching, both used for absolute identification, along with eyewitness testimony – none are perfect, yet all play vital roles in our justice system. Even human facial recognition ability, which we rely on daily, isn't infallible. Studies have shown that eyewitnesses can misidentify suspects, particularly across racial lines or in high-stress situations. Yet we don't dismiss human recognition altogether; we supplement it with other evidence and expertise.
This is precisely how we should approach facial recognition technology. Rather than seeking perfect accuracy – an impossible standard that we don't apply to other investigative methods – we should leverage its strengths while acknowledging its limitations. The technology excels at rapidly comparing images against large databases of mugshots. This is a task that would take human investigators countless hours to complete.
A practical and balanced approach would use facial recognition to generate leads rather than conclusions. When investigating a crime, law enforcement could use the technology to create a shortlist of potential matches from their database. These candidates would then undergo thorough human review by trained officers who can apply their expertise and judgment. This AI-human partnership combines the speed and scale of technology with the nuanced understanding that only human investigators can provide.
Moreover, facial recognition is far from a novel or experimental technology. Its foundations lie in decades of research across multiple scientific disciplines, including computer vision, pattern recognition and machine learning. Major academic institutions, technology companies and government agencies have invested heavily in its development and validation. The scientific community has extensively peer-reviewed its underlying principles through countless research papers and conferences.
Critics often focus on two types of errors: false positives and false negatives. While false negatives (failing to identify a present match) might mean missed opportunities, they don't risk wrongful accusations. False positives, however, require careful handling. This is precisely where human oversight becomes crucial. By treating facial recognition matches as investigative leads rather than definitive evidence, we create a system of checks and balances that mitigates the risk of false positives while capitalizing on the technology's capabilities.
The path forward is clear: we should embrace facial recognition technology as an investigative aid while maintaining appropriate boundaries. This means using it to generate leads and support investigations, but not as standalone evidence for positive identification in court. By combining technological capabilities with human expertise, we create a more effective investigative process that serves justice while protecting individual rights.
The question isn't whether to use facial recognition technology, but how to use it responsibly. As our technological capabilities continue to advance, our legal system must evolve to accommodate these tools while maintaining its commitment to justice and fairness. A balanced approach that leverages both artificial and human intelligence offers the best path forward.
Dr. Manjeet Rege is a distinguished academic and industry leader in the fields of data science and artificial intelligence. As a professor and the chair of the Department of Software Engineering and Data Science at the University of St. Thomas, he has made substantial contributions to the academic world, evidenced by his recognition as a Leading Academic Data Leader for 2023 by CDO Magazine. Rege also serves as the director of the Center for Applied Artificial Intelligence at the University of St. Thomas, where he oversees initiatives that blend academic research with practical applications in AI.
