As a criminologist, Renee Cummings said she has seen first-hand and "in the blink of an eye" how an algorithm can change the trajectory of someone's life, oftentimes in a way that is negative. It's ultimately that work that led her to her current research as an instructor, AI ethicist and data activist at the University of Virginia.
Cummings shared some of the findings of her work as the keynote speaker at the Partnership to Advance Responsible Technology's (PART) inaugural summit event held at the Phipps Conservatory in Oakland on Tuesday. A Pittsburgh-based nonprofit think-tank, PART strives to engage stakeholders across industries to ensure emerging and data-driven technologies, especially AI, are researched, developed, deployed and governed responsibly. PART released its first foundation-supported research report titled "How Pittsburgh Can Build a Better Innovation and Emerging-Technology Economy Through Connectivity, Density, and Collaboration" back in January.
For Cummings, a lot of her research focuses on the building of algorithms and the use of AI within them. But when it comes to building these algorithms, Cummings argued that the potential risks that come with them must be thought of ahead of any implementation. After all, Cummings said, it's easier to fix a technological problem before it's deployed at scale than it is to then fix it once it's begun.
But there's also much more to consider beyond potential business-related ramifications when it comes to the deployment of AI, Cummings said.
"I spend a lot of time in the C-suite speaking a lot with executives … but when I go into the C-suite it's usually because there's a crisis, and the crisis could create — because of an algorithm — a criminal investigation, of course then diminish public trust, a media attack, regulatory challenges, reputational damage and, of course, revenue loss," Cummings said. "These are the ethical risks. We've always found from a business perspective that we are more comfortable with the ethical risks but you've got to not only think about the risk approach to big tech or the risk approach to responsible tech or the risk approach to AI. We've got to think about that rights approach because it's about risks but it's also about rights."
Cummings noted harmful uses of algorithms by businesses that have taken human-based biases and used them at scale to affect minority populations. She specifically called out algorithms that have been used to establish credit worthiness, loan and mortgage approvals and digital redlining — the practice of making and maintaining inequities between marginalized populations with digital technologies — all of which have in some capacities, Cummings said, prevented marginalized people from equal access to opportunities or services.
Those in themselves are risks, she said, and they're risks in need of addressing as AI becomes even more prevalent across industries.
"When we're thinking about those risks, we're thinking about those risks within the context of the politics and the power and the privilege and the prejudice and the fact that things — like accessibility and voice and visibility, identity, representation — are so critical and are so easily undermined in the blink of an eye with the deployment of an algorithm," Cummings said. "When we think about the ethical risk, these are the basic ones that we (must) always discuss."
In her closing remarks, Cummings called on the more than 75 people in the room — some of them being data scientists, researchers, academics and policymakers — to use their power to use technology properly and to see to it in a just and equitable way.
"Doing technology right is really about finding ways to harness the power of this technology and to simply reduce the harm," Cummings said. "And if we can harness the power of this technology, then we can really build an extraordinary world where that world would really be able to create the kinds of opportunities that are required to create legacies."
For Lance Lindauer, executive director and co-founder of PART, Cumming's speech perfectly capped a day-long event filled with discussions on how the region can better lead a global discussion on responsible uses of emerging technologies.
"I think the biggest thing that I took away from today was everybody buying into the notion that this is a larger conversation than what traditional tech narratives are like," Lindauer said. "It's policy-related, it's education-related, it's financial investment-related from all these different sectors, and so hearing that so many folks had questions and so many folks were having really robust and loud conversations over the breaks and the lunch that everybody assembled here today really bought into, 'hey, this involves me too, and this involves my industry, and I'm not a CEO, but it involves every layer of every job that folks are working on.'"
