Last week, Fourth View hosted a Zoom webinar titled, “Tech Toks: Is AI Racist?” with University of Michigan Professor and Director of the Michigan Institute for Data Science (MIDAS) H.V. Jagadish, Executive Director of the Surveillance Technology Oversight Project (S.T.O.P.) Albert Fox Cahn, and Director of Civil Liberties at TechFreedom Ashkhen Kazaryan. It explored how technology is intersecting with race. From processing college applications to making lending decisions, the conventional wisdom that a computer would be more equitable may not necessarily be the case.
H.V. Jagadish gave a clear explanation as to what an algorithm was, the difference between the categories of “AI” and “algorithms,” and some of its capabilities, but made a cautionary note that AI needs to be monitored on a situational basis. And while AI can be very efficient and accurate, he acknowledged that the data it’s fed can be corrupted by historical bias. He did say, though, that the algorithms can be fixed and improved. He recommended more testing needs to be conducted by institutions that rely on AI to improve them and there needs to be accountability by others outside the system to check them.
Ash gave a synopsis of policies over time that have interacted with technology and the current state of policies, highlighting the lack of understanding by legislators when it comes to technology in general. This illustrated how underdeveloped and lacking current or any future legislation has been on the matter.
Albert laid out the current groundwork for lawyers and how he was skeptical as to the progress the tech industry has made in bettering the fairness of its AI. Citing multiple examples of vicious feedback loops, such as criminal sentencing or law enforcement’s predictive policing models, he wasn’t satisfied with leaving it up to tech companies to regulate themselves. Some federal legislation may help, but beefing up legal curriculums with tech related classes is a good start.