[Editor’s Note: This post was written by CRA’s new Tisdale Policy Fellow for Summer 2019, Jesse Anderson.]
On Wednesday, June 26 the House Science, Space, and Technology Committee held a hearing titled Artificial Intelligence: Societal and Ethical Implications to review the diverse ethical and social implications of Artificial Intelligence (AI). The committee heard from panelists about the complications of AI as well as several policy recommendations.
Chairwoman Eddie Bernice Johnson (D-TX) emphasized the need for more conscious involvement of ethics at every stage of the AI research and development process, noting that, “while a few individual agencies are making ethics a priority, the Administration’s executive order and strategic plan fall short in that regard.” Chairwoman Johnson noted that, though ethics have been considered in the creation of AI technology, they are largely seen, “as an add-on rather than an integral component of all AI R&D.” Johnson called for more investment in AI R&D to maintain the nation’s global dominance in the field. Johnson noted that more investment in the field is critical for continued US leadership in AI, and that leadership will have the added benefit of instilling American values into AI systems. Jim Baird (R-Indiana) reiterated Johnson’s call for more investment toward resources for researchers. He underscored the need for investment in the field to sustain America’s superiority in AI R&D.
The witnesses before the committee were Meredith Whittaker, Co-Founder, AI Now Institute, New York University; Jack Clark, Policy Director, OpenAI; Joy Buolamwini, Founder, Algorithmic Justice League; and Georgia Tourassi, Director, Health Data Sciences Institute, Oak Ridge National Laboratory. They illuminated several of the ethical challenges that AI presents and suggested some policy solutions. Among the solutions discussed, many of the panelists emphasized the need for interdisciplinary research in AI and the direction of more advanced resources to researchers.
The panelists spotlighted the issue of bias encoding in AI technology, where innovators code human biases into the technology. Whittaker noted several instances of gender discrimination – such as studies that show that, “voice recognition hears masculine sounding voices better than feminine sounding voices.” Clark agreed, saying that the need for value judgments when programming AI can lead to the passage of biased values. The concentration of AI institutions, Whittaker argued, helps companies purvey marketing materials that only explain the benefits of AI. She also noted that many of these companies are “non-diverse.” Whittaker, like Johnson, said that further research necessitates more investment. This investment would go toward resources that researchers would otherwise have to access through employment at large technology companies. These resources would allow researchers to reinforce America’s technological prowess in the AI field. Clark also suggested increased funding for interdisciplinary research, arguing that bringing together diverse professionals will help combat the bias issues. Clark called for the government to convene with academia and industry to devise more effective policy. He too underscored the importance of broadening the scope of resources available to researchers.
Buolamwini criticized the lack of transparency in the tech industry as well as a reliance on poor data. Like other panelists, she provided real-life examples of AI’s implications. Buolamwini called attention to a June 2019 study that, “showed that for the task of pedestrian tracking children were less likely to be detected than adults.” Like Clark, she emphasized the need for more funding toward AI research. This government-funded research on human centered AI needs to be interdisciplinary and inclusive, as Buolamwini noted that marginalized demographics bear the brunt of AI’s negative implications.
As Johnson indicates, “leadership is not just about advancing the technology, it’s about advancing it responsibly.” The committee widely agreed that a focus on interdisciplinary research, further investment in resources for researchers, and the retention of global dominance in the AI field will be critical aspects of future policy solutions. AI will continue to be a significant policy concern for Congress and the Federal government as a whole. CRA will continue to monitor developments that arise in government and policy circles; be sure to check back for new updates.