TL;DR Breakdown
- Rumman Chowdhury believes that ethical AI surveillance is not feasible, stating, “We cannot put lipstick on a pig.’’
- She advocates for the establishment of an external global governance board to oversee AI technologies and address issues like surveillance capitalism.
- Chowdhury emphasizes the importance of accountability and collaboration in governing AI, cautioning against over-regulation while addressing ethical challenges.
In a recent interview, Rumman Chowdhury, former director of machine learning ethics, transparency, and accountability at Twitter, expressed her skepticism regarding the possibility of ethical artificial intelligence (AI) surveillance. Speaking at New York University’s School of Social Sciences, Chowdhury stated, “We cannot put lipstick on a pig. I do not think ethical surveillance can exist.”
Chowdhury, known for her work in AI ethics, highlighted her concerns about the rise of surveillance capitalism, which she believes is inherently racist and should not be supported. In a Wired op-ed, she advocated for the establishment of an external global governance board to oversee AI technologies, addressing issues such as surveillance capitalism, job disruption, and nuclear misinformation. Chowdhury emphasized the need for a diverse board that is not tied to any single institution and represents a global perspective.
Accountability and regulation challenges in the AI industry
According to Chowdhury, accountability is a crucial aspect that is often lacking in the AI industry. She pointed out that internal risk analysis within companies tends to overlook moral considerations. When the risk of failure or reputational harm becomes significant, she noted, organizations often bend the rules in their favor. Chowdhury believes that addressing these challenges requires a collective effort, involving various stakeholders and cooperation in a global scale.
While acknowledging that regulation is necessary, Chowdhury cautioned against over-regulation, which could lead to models overcorrecting and failing to address underlying issues. She emphasized the complex nature of defining what is toxic or hateful, stating that it is an ongoing journey that requires continuous evaluation and adaptation.
Rumman Chowdhury believes technology should not divorce itself from humanity
Chowdhury’s work aims to bridge the gap between technologists and policymakers, advocating for a better understanding of the human impact of technology. She believes that technology should not divorce itself from humanity but should instead be harnessed to address human flaws and improve society.
Rumman Chowdhury, a prominent figure in the field of AI ethics, has always been intrigued by the complexity and unpredictability of human behavior. During her undergraduate studies in political science at MIT, she realized the potential for using models and data more effectively in non-profit organizations. This led her to pursue a master’s degree in quantitative methods at Columbia University, where she further honed her skills.
Chowdhury’s work has taken her to various locations around the world in the past month alone. From Spain to San Francisco, Boston to New York, and upcoming talks in Vienna and Pittsburgh, she is actively engaged in addressing pressing issues related to AI, such as cybersecurity, nuclear misinformation, and language learning platforms like Duolingo.
Leverage incentive structures and decentralize power.
At the core of Chowdhury’s approach lies a simple yet powerful dictum: listen, communicate, collaborate. However, she recognizes that even when industry leaders express their commitment to preventing AI harm, there can often be a discrepancy between words and actions. She observes that during times of heightened scrutiny, some entities may call for regulation while simultaneously employing lobbying tactics to hinder the passing of regulatory laws.
Chowdhury’s approach to AI ethics and governance is grounded in the understanding that people, unlike machines, have diverse motivations and priorities. Effecting meaningful change requires leveraging incentive structures and decentralizing power. She challenges the notion of a single hero narrative and emphasizes the importance of collaboration and collective decision-making in shaping the future of AI.
