Research initiatives

In the United States, many jurisdictions are turning to pretrial risk assessment tools as an alternative to cash bail. More than 60 risk assessment tools are in use throughout the country today, at both the federal and state level and in multiple stages of a person’s process through the carceral system. In what follows, we analyze risk assessment tools for their claims of providing objective, empirical evidence to inform decisions around pretrial release and detention. We find serious concerns with their ability to generate accurate, valid, reliable, and unbiased outcomes and to address root...

Automated License Plate Readers (ALPRs) are a surveillance technology that can alert law enforcement about vehicle locations in real-time or provide information on past movements. In recent years, growing numbers of public and private entities have begun using ALPRs, moving some communities to implement policies aimed at limiting the potential damage posed by the technology. This memo analyzes legal and policy concerns related to the technology. It concludes by suggesting policy options that communities can advocate for, including a ban or moratorium and appropriate safeguards if the...

Decentralized wastewater treatment is the future of wastewater treatment. In developing countries without sophisticated wastewater infrastructure, it is the design of choice since it has lower capital costs, protects the local environment, and incentivizes water re-use. Though promising, technological limitations and ‘cultural inertia’ are currently the main drawback of decentralized wastewater treatment in developed countries. Centralized wastewater plants are the predominant technology in the U.S. Implementing decentralized treatment lowers the value of the existing infrastructure and...

Acoustic gun detection systems (AGDS) are a law enforcement technology designed to detect gunshot sounds and notify police of the event and location in close to real time. ShotSpotter is the most well-known AGDS in the United States, sold by a company of the same name. In what follows, we analyze ShotSpotter and the company’s claims that the system improves safety, reduces crime, and promotes positive relationships between communities and police. We conclude that the technology’s accuracy, effectiveness, cost, and systemic biases raise serious concerns. Communities concerned with local law...

Large language models (LLMs)—machine learning algorithms that can recognize, summarize, translate, predict, and generate human languages on the basis of very large text-based datasets—are likely to provide the most convincing computer-generated imitation of human language yet. Because language generated by LLMs will be more sophisticated and human-like than their predecessors, and because they perform better on tasks for which they have not been explicitly trained, we expect that they will be widely used. Policymakers might use them to assess public sentiment about pending legislation,...

STPP is collaborating with the College of Engineering, Michigan Institute for Data Science, and the Detroit Community Technology Project to reimagine the undergraduate computer science curriculum to include sustained attention to social, moral, equity, and policy dimensions of data and technology. The funded project focuses on both undergraduate and postdoctoral training in public interest technology. The funded project focuses on both undergraduate and postdoctoral training in public interest technology. Postdoctoral fellow Johanna Okerlund received training in the equity, justice, and...

In winter 2020, a novel coronavirus (SARSCoV-2) that caused COVID-19 started its spread across the globe, and by July 2020, over 500,000 people worldwide had died of the disease. By March 2021, there were over 120 million cases and over 2.8 million deaths. To combat the pandemic and return to “normalcy”, experts estimate that at least 80% of the world’s population needs to be resistant to the virus, and most of the world’s population will require vaccination. This will be a challenge. In addition to facilitating widespread distribution, governments will need to combat “vaccine hesitancy”: an...

Facial recognition (FR) technology was long considered science fiction, but it is now part of everyday life for people all over the world. FR systems identify or verify an individual’s identity based on a digitized image alone, and are commonly used for identity verification, security, and surveillance in a variety of settings including law enforcement, commerce, and transportation. Schools have also begun to use it to track students and visitors for a range of uses, from automating attendance to school security. FR can be used to identify people in photos, videos, and in real time, and is...