REGULATING FACEBOOK An ethical analysis of AI run amok, political posturing, and the failure of self-regulation

Abstract

The burgeoning field of AI ethics posits that AI is not just another utility that can be regulated once matured, but a powerful new form of smart agency that must be steered from the onset to benefit society.1 But what happens when the proverbial train has already left the station? What are the challenges of regulating an AI application after its transformative power has been fully unleashed on society? The purpose of this paper is to explore the challenges with ex parte regulation of AI, using one of its most notorious examples, Facebook, as a detailed case study. It demonstrates that regulation cannot be left up to the free market, and that policy makers must be more proactive in setting the parameters for the ethical development of artificial intelligence. While much has been written about the negative externalities of Facebook and other social media platforms, this paper offers a unique perspective by using the prism of AI ethics to evaluate how American political actors have responded to the Facebook controversy and to assess the company’s efforts to self-regulate. The paper uses a seminal moment in the company’s history to explore ethical questions around the regulation of Facebook algorithms: CEO Mark Zuckerberg’s appearance in extraordinary full day hearings before the U.S. Senate and House of Representatives in April 2018 to answer for his company’s role in the Cambridge Analytica and Russian election interference scandals. Using five guiding principles (beneficence, non-maleficence, autonomy, justice and explicability) from the field of AI ethics as an analytical lens, a qualitative textual analysis of the legislative hearings was conducted to examine how issues were framed by Zuckerberg and U.S. legislators, how they were addressed prospectively, and what issues were missed from a public policy perspective. Policy considerations are further examined by evaluating Facebook’s efforts to self-regulate in response to the hearings, and exploring potential pathways for regulators. The paper concludes that while legislators were able to identify ethical challenges with Facebook algorithms in alignment with the five principles, effective regulatory proposals were hampered by a polarized congress, conflicting interpretations of the issues along partisan lines and a shallow understanding of the technology. It also shows that while self-regulation allows for more immediate corrective action, efforts taken by Facebook to date have been inconsistent, insufficient, and motivated by profit over ethics. Finally, this paper makes the case for better ex ante regulation of new AI technology that is not subject to the whims of political parties, with the need for clear legislative and regulatory boundaries, and independent oversight to protect the health of individuals, society, and democracy. In an era of rapid technological change that is expected to lead to ever greater societal disruptions, the Facebook story serves as an important cautionary tale on the need for a more proactive approach to policy-making.