Pitt cyber institute starts task force to study big data tools used by city, county

A new collaboration spearheaded by the University of Pittsburgh’s cyber institute aims to help governments address the challenge of minimizing bias in data tools using algorithms and artificial intelligence.

The newly formed Pittsburgh Task Force on Public Algorithms was announced Wednesday morning and includes participation from Pittsburgh and Allegheny County public agencies. Its work will span two years, culminating with a report spotlighting problems and establishing best practices moving forward.

“We have begun to make a lot of decisions relying on big data,” former U.S. Attorney David Hickton, founder and director of the University of Pittsburgh’s Institute for Cyber Law, Policy and Security, told the Tribune-Review. “We do that because we have the data available, we’re in the internet age and it’s faster and cheaper to use the tools — and the aspiration is that it will also be better.”

Big data tools increasingly are driving key decisions by Western Pennsylvania’s public agencies, from law enforcement to the courts to child welfare offices.

In Allegheny County, algorithms are being used by call screeners to identify children at risk of being abused and by judges to help set bail amounts and probation sentences.

“We’re doing the best we know how, but we want to be responsive to the community,” said Erin Dalton, deputy director of the Allegheny County Department of Human Services..

Dalton will be on the task force’s government advisory panel.

“I see it as an enabler of our work. I think this group is thoughtful about wanting government to innovate, be accountable and responsible to the public,” Dalton said.

In Pittsburgh, police are analyzing data to determine where to deploy officers and gauge how dangerous a suspect or domestic violence incident could be.

Shatara Murphy, assistant director of community affairs for the Pittsburgh Public Safety department, said she’s excited for the group to begin work and is happy to be a part of the advisory panel. The group meets for the first time next week.

Government uses data and algorithms to make informed decisions, and the task force will add a human check to how that data is collected and used, Murphy said.

It will be a step toward “holding technology accountable,” Murphy said.

The task force can also help communicate how data is used to residents who may not otherwise know, she said.

“The community component is the thing I’m most excited about,” she said.

Hickton said the potential dilemma is that automated data systems designed by people are not without inherent biases. He said the availability of digital tools has expanded so rapidly that few public agencies have scrutinized whether their systems are ethical, fair and effective. It’s akin to “flying the plane before we built it,” he said.

“If you don’t take the steps required to make sure the algorithms are neutral, what can happen is that existing patterns of inequality by race or economics can be made worse,” Hickton said.

Hickton emphasized that he’s not opposed to governments turning to data to inform decisions. He lauds agencies for leveraging data to improve their work and maximize time and resources.

Hickton said that without close scrutiny and diligent improvements being made to the data sources and algorithms in place, such efforts could prove ineffective. Or worse, they could inadvertently cause harm in the form of exacerbating persisting inequities.

“Whether you’re trying to identify which kids are at risk, what you can do in preventive policing, how we can get through airports faster and how we can find terrorists, whether you’re talking about artificial intelligence or facial recognition technology or decision making algorithms, they’re all only as good as their underlying data,” he said.

As an example, Hickton cautioned against sending police to a particular location because historical data shows a large amounts of arrests or police activities. That data could reflect a pattern of over-policing black neighborhoods, Hickton said.

Problems with data sources and systems also can produce skewed results that could allocate resources incorrectly or unfairly disburse them, Hickton said.

Hickton will chair the 22-member task force, which includes nonprofit leaders, university researchers, community service providers and legal and digital science experts. They’ll have input from an advisory panel of digital services employees from Allegheny County and the city of Pittsburgh.

“Particularly, as governments expand the use of these powerful tools, with so much at stake for our liberty, the fight for algorithmic fairness and accountability is another frontier in the struggle for civil rights,” said task force member Esther Bush, president of the Urban League of Greater Pittsburgh, which aims to improve the lives of blacks across the region.

The Heinz Endowments is providing grants to fund the two-year effort, which also has support from Common Cause Consultants and Weiss Burkhardt Kramer, LLC.

“As Pittsburgh develops into one of the world’s leading centers of research and deployment of artificial intelligence, machine learning, and other emerging technologies, it is imperative that we simultaneously develop a set of ethics, policies, and procedures informed by people who will be impacted by these technologies,” said Carmen Anderson, chief equity officer for The Heinz Endowments.

The public will have the chance to learn more and share individual concerns at two community outreach meetings scheduled for March, including one in Homewood. The task force aims to finish its work and publish its final report in summer 2021.

Natasha Lindstrom and Tom Davidson are Tribune-Review staff writers. You can contact Natasha at 412-380-8514, nlindstrom@tribweb.com or via Twitter @NewsNatasha. You can contact Tom at 724-226-4715, tdavidson@tribweb.com or via Twitter @TribDavidson.