The Human Rights Data Analysis Group, an independent scientific human rights organization based in San Francisco, has counted at least 17,723 people killed in Syrian custody from 2011 to 2015 — around 300 every week — almost certainly a vast undercount, it says.

According to the Human Rights Data Analysis Group, at least 17,723 people were killed in government custody from the start of the uprising in March 2011 to December 2015 – an average of 300 deaths each month. There are no figures for subsequent years but there is no reason to believe the killings stopped.

The Human Rights Data Analysis Group employs AI to analyze data from conflict zones, identifying patterns of human rights abuses that might be overlooked. This assists international organizations in holding perpetrators accountable.

HRDAG is mentioned in the “child welfare (sometimes called “family policing”)” section: At least 72,000 low-income children are exposed to AI-related decision-making through government child welfare agencies’ use of AI to determine if they are likely to be neglected. As a result, these children experience heightened risk of being separated from their parents and placed in foster care.

“Patrick Ball, a statistician at the US-based Human Rights Data Analysis Group not involved in the research, has used capture-recapture methods to estimate death tolls for conflicts in Guatemala, Kosovo, Peru and Colombia.

Ball told AFP the well-tested technique has been used for centuries and that the researchers had reached “a good estimate” for Gaza.”

“Patrick Ball, a statistician at the US-based Human Rights Data Analysis Group not involved in the research, has used capture-recapture methods to estimate death tolls for conflicts in Guatemala, Kosovo, Peru and Colombia.

Ball told AFP the well-tested technique had been used for centuries and that the researchers had reached “a good estimate” for Gaza.”

Patrick Ball, director of research at the Human Rights Data Analysis Group, and a statistician who has conducted similar estimates of violent deaths in conflicts in other regions, said the study was strong and well reasoned. But he cautioned that the authors may have underestimated the amount of uncertainty caused by the ongoing conflict.

The authors used different variations of mathematical models in their calculations, but Dr. Ball said that rather than presenting a single figure — 64,260 deaths — as the estimate, it may have been more appropriate to present the number of deaths as a range from 47,457 to 88,332 deaths, a span that encompasses all of the estimates produced by modeling the overlap among the three lists.

“It’s really hard to do this kind of thing in the middle of a conflict,” Dr. Ball said. “It takes time, and it takes access. I think you could say the range is larger, and that would be plausible.”

Eric Sears, a director at the MacArthur Foundation who leads the grantmaker’s Technology in the Public Interest program, worked at Human Rights First and Amnesty International before joining MacArthur, and has been following HRDAG’s work for years. … One of HRDAG’s strengths is the long relationships it maintains with partners around the globe. “HRDAG is notable in that it really develops deep relationships and partnerships and trust with organizations and actors in different parts of the world,” Sears said. “I think they’re unique in the sense that they don’t parachute into a situation and do a project and leave. They tend to stick with organizations and with issues over the long term, and continually help build cases around evidence and documentation to ensure that when the day comes, when accountability is possible, the facts and the evidence are there.”

“Kristian Lum… organized a workshop together with Elizabeth Bender, a staff attorney for the NY Legal Aid Society and former public defender, and Terrence Wilkerson, an innocent man who had been arrested and could not afford bail. Together, they shared first hand experience about the obstacles and inefficiencies that occur in the legal system, providing valuable context to the debate around COMPAS.”

Kristian Lum is focusing on artificial intelligence and the controversial use of predictive policing and sentencing programs.

What’s the relationship between statistics and AI and machine learning?

AI seems to be a sort of catchall for predictive modeling and computer modeling. There was this great tweet that said something like, “It’s AI when you’re trying to raise money, ML when you’re trying to hire developers, and statistics when you’re actually doing it.” I thought that was pretty accurate.

NPR Illinois has covered the new National Police Index, created by HRDAG’s Tarak Shah, Ayyub Ibrahim of Innocence Project, and Sam Stecklow of Invisible Institute.

“If you’re not careful, you risk automating the exact same biases these programs are supposed to eliminate,” says Kristian Lum, the lead statistician at the San Francisco-based, non-profit Human Rights Data Analysis Group (HRDAG). Last year, Lum and a co-author showed that PredPol, a program for police departments that predicts hotspots where future crime might occur, could potentially get stuck in a feedback loop of over-policing majority black and brown neighbourhoods. The program was “learning” from previous crime reports. For Samuel Sinyangwe, a justice activist and policy researcher, this kind of approach is “especially nefarious” because police can say: “We’re not being biased, we’re just doing what the math tells us.” And the public perception might be that the algorithms are impartial.

When Patrick Ball was introduced to Ibero’s database, the director of research at the Human Rights Data Analysis Group in San Francisco, California, saw an opportunity to turn the data into a predictive model. Ball, who has used similar models to document human rights violations from Syria to Guatemala, soon invited Data Cívica, a Mexico City–based nonprofit that creates tools for analyzing data, to join the project.

“Every kind of classification system – human or machine – has several kinds of errors it might make,” [Patrick Ball] says. “To frame that in a machine learning context, what kind of error do we want the machine to make?” HRDAG’s work on predictive policing shows that “predictive policing” finds patterns in police records, not patterns in occurrence of crime.

Kristian Lum discusses the challenges of getting accurate data from conflict zones, as well as her concerns about predictive policing if law enforcement gets it wrong.

Work by HRDAG researchers Kristian Lum and William Isaac is cited in this article about the Policing Project: “While this bias knows no color or socioeconomic class, Lum and her HRDAG colleague William Isaac demonstrate that it can lead to policing that unfairly targets minorities and those living in poorer neighborhoods.”

Kristian Lum’s work on the HRDAG Policing Project is referred to here: “In fact, Lum argues, it’s not clear how well this model worked at depicting the situation in Oakland. Those data on drug crimes were biased, she now reports. The problem was not deliberate, she says. Rather, data collectors just missed some criminals and crime sites. So data on them never made it into her model.”