December 1, 2020

Human Contexts and Ethics

It somehow seems appropriate that courses for UC Berkeley’s Human Contexts and Ethics (HCE) program are being taught online via Zoom.

With 500 or so students each in their virtual box, coming together as a community to think about the role of technology in shaping how they think about themselves and their options for understanding and using data science to shape their world, they are participating in a real-time experiment around the program’s objectives.

“We want to help our students gain a socio-technical perspective -- to look around at the world we live in where data plays such a core role and see how people have shaped it through decades and centuries of social and technical forces,” said Margo Boenig-Liptsin, director of the HCE program and co-instructor of Data 104: Human Context and Ethics of Data. “This perspective enables students to reflect on how their actions as data scientists, consumers, and citizens in the datafied world contribute to shaping collective life.”

One class offered in HCE in the Fall 2020 semester is Data 104: Human Contexts and Ethics of Data, which starts with the premise that as data-driven services and artificial intelligence-powered devices shape the many aspects of our lives, “they are products of historical and institutional dynamics, deep-rooted social structures and political cultures that bear the marks of human intentions, interests, and desires.” In Spring 2021, in addition to Data 104, a new offering will be Data 4AC -- Data and Justice, which will engage students with fundamental questions of justice and race in relation to data and computing in American society.

Data 104 was first taught in 2018 and is one of the backbone signature data science courses in Data Science Undergraduate Studies at Berkeley, alongside Data 8, Data 100, and Data 102 as part of the Data Science Education Program and is now within the Division of Computing, Data Science, and Society. But unlike many of the other classes taught by campus units in the division, Data 104 is presented through the History Department. Boenig-Liptsin, Edmundson and HCE founder Cathryn Carson all have Ph.D.s in history. The courses teach students “to use thetools of applied historical thinking and Science, Technology, and Society to recognize, analyze, and shape the human contexts and ethics of data.”

In Data 104, Boenig-Liptsin and Edmundson use examples from the past and present to show how humans have always used technologies in the service of building social orders. Scenarios include the rise of statistics with the modern nation-state in the 18th century, racial classification techniques of South Africans under Apartheid, and pre-digital examples of technologies, such as streetlights and bridges, to which people delegate power to organize human affairs.

“New technologies have always been part of the perennial systems of humans and human dynamics, which are also transformed by those technologies,” Boenig-Liptsin said.

“We teach our students ways of thinking about issues like power, agency, and materiality from the interpretive social sciences. For example, how does making knowledge with open source data science tools reconfigure power dynamics of a community? What happens to human agency and responsibility when actions are delegated to technologies? How does the seemingly immaterial metaphor of "cloud" computing connect up with the intensive land, electricity, and water resources necessary to maintain this crucial infrastructure?”

This often means changing students’ perceptions of such concepts as ethics and privacy.

“Frequently students begin the semester with a narrow notion of ethics as it applies to data science, such as concerns about privacy, fairness and transparency and they might look to us for an ethical code to follow,” Boenig-Liptsin said. “They’re surprised by how these values have histories and have been shaped by human desires, laws, and the history of computing.”

History allows students to not only distinguish the patterns of the past in order to know what to watch for in the future, but to also recognize the role of human actions in the making of the present, according to Boenig-Liptsin. “Once they can deconstruct how we got to the present, they can focus on what ethical action means today, that is, how to aim at human flourishing in collectives and in the specific realities of the datafied world," she said. 

For example, users of online technology are often unaware that they are part of ongoing experiments. An online platform may subtly shift when viewed by different users, showing similar but different images and text to see how that affects behaviors. The information is then used to refine the platform to move users toward specific outcomes. Everything is in flux, which is why the instructors aren’t trying to teach the students to do one thing and not another.

“A lot of our students go through discomfort when they think about living in a world filled with ambiguity,” said Edmundson, who has studied philosophy and critical theory. “They have a lot of questions about the ambiguities of human situations and that they are not definable and there is no clear solution. Though it’s uncomfortable, we encourage them to embrace it and see it as a source of power, not paralysis.”

Edmundson sees this power as the intersection of politics, personal power and data science.

“We try to convey a sense of publicness, of unity and togetherness and have the students think of opportunities for creating new social movements,” he said. 

But to do so, there has to be an underlying structure. As an example, after the first Black Lives Matter protests began in July 2013 following the acquittal of Trayvon Martin’s murderer in Florida, the movement was localized and no central organizing structure was created. It was reborn in 2020 with the murder of George Floyd, Ahmaud Arbery, and Breonna Taylor and led to the creation of a political action committee in October.

Race was central in Data 104 when it was first taught, as the instructors wanted to demonstrate differences in how groups of people have or don’t have access to power is often based on society’s construction of an individual, such as through racial profiling, which can determine who “belongs” and who doesn’t. 

“This was not part of the standard data science conversation,” Boenig-Liptsin said. “By giving students a more empowered view, they can be part of reconstructing society not by micro-decisions but by contributing to making a new collective reality. We give them the tools that can be used to improve existing systems.”

The courses also aim to help students learn how they can work with other people, which is essential to making changes. “When you work among people who see the world differently, and you listen to them, you will change yourself,” Boenig-Liptsin said. 

“Social progress is not going to come through new technologies alone, but through new communities,” Edmundson said. “Large-scale changes, such as ongoing demonstrations or disruptions like the COVID pandemic present opportunities for creating new social movements.”

And having a better understanding of data science is a powerful tool for understanding issues and driving change, whether fighting discriminatory gerrymandering of voting districts, voting on a state initiative that purports to increase privacy rights or claims to improve the lives of gig workers, or determining which news is real and which is fake.

“Increasing oversight of current systems requires new structures and new ways for the public to take part and weigh in,” Boenig-Liptsin said. “We need new checks and balances, not just more regulation. We need to recognize the need for social innovation.”

To encourage students to see how all of these ideas come together and how they can effect change, Data 104 includes a capstone project called the HCE Vignette. In it, the students position themselves as an expert in a specific field and describe how they would reach out to a public and participate in social engagement and change. The idea is for them to see themselves as burgeoning experts while also recognizing their limitations, Edmundson said.

How it all got started

“No other university, no other place in the world, has an HCE program at the scale of Berkeley,” said Cathryn Carson, who holds the Thomas M. Siebel Presidential Chair in the History of Science at Berkeley and started the HCE program. “It was built into the Data Science Education Program from the start, not as an add-on.” 

But Carson didn’t start her career with such a program in mind. She originally considered a career in theoretical physics, but many career options for that path led to defense work, which wasn’t what she wanted to put her life toward. When the Cold War ended, she decided to become a historian of physics, “which is so much the discipline of the 20th Century,” she said. She earned her master’s in physics in 1993 and Ph.D. in history of science in 1995, both from Harvard University. This strong basis in the history of science and technology, and how these areas affect society, helped form the HCE program and continues to provide context for the classes. 

“As data science is now shaping the 21st Century, we are seeing how we can bring what we learned in the last century into the conscious framing of how Berkeley teaches data science,” Carson said. “How do we integrate all of the knowledge we have here on the HCE team--each of us brings our own deep disciplinary grounding, and we have a sense of where we need to be shaping things before they appear.

“There is no road map, but we bring our own compasses,” Carson said.

“And we’re doing it in real-time, listening to our students and informed by an intuitive and disciplined sense of what matters regarding the human condition,” Boenig-Liptsin said.