Richard Olaniyan
According to a recent estimate, Smart Homes should comprise about 53.9% of U.S. homes by 2023 (Smart Home Report, 2019). Voice-Activated Personal Assistants (VAPAs)—Amazon’s Alexa, Apple’s Siri and Google Assistant—are also becoming commonplace in an “ideal” Smart Home. At the same rate, concerns about user privacy—both within the Smart Home space and generally in the tech industry—has kept up pace (Shields, 2018). For example, data stored by VAPAs has been misconstrued and used by law enforcement (Zwerdling, 2018), and reports have surfaced of VAPAs eavesdropping on users (Newman, 2018). Although the world has benefited positively from the advancement of these technologies, advancements in AI and Machine Learning technologies may also have drawbacks.
"The results from this study will give policymakers and tech companies a peek into society’s ethical expectations as they relate to the moral decisions of VAPAs."
Speculative design, popularized by Dunne and Raby (2013), puts forward the notion of asking “what-if” questions to probe an expected future now. With a combination of design fiction, stories, and scenarios, we can essentially ask deeper and pointed questions about the potential role of technologies we currently use. For example, in 2018, researchers at the Massachusetts Institute of Technology (MIT) asked questions about the potential ethical dilemmas that will face autonomous vehicles. Using an online experimental platform called the Moral Machine, they were able to capture “societal expectations about the ethical principles that should guide machine behaviors” (Awad et al., 2018). With millions of respondents from more than 233 countries, MIT researchers concluded that it might be impossible to create a set of guidelines that informs our understanding of machine ethics due to strong cultural variations in the result. However, they found that there was consensus along certain themes like the “preference for sparing young lives and more lives” (Awad et al., 2018).
This paper guides a similar conversation around VAPAs. Using an experiment delivered through an online survey, participants were placed in the “judge mode procedure” adapted from Awad et al. (2018). For this exercise, three different scenarios, all having potential implications for law enforcement, were displayed to participants in the Judge Mode. In each scenario, users are prompted to choose an outcome which indicate whether the VAPA should carry out an action or not. Each scenario came with a detailed storyline to which users were told to respond to the question: “What should Morli (the Voice-Activated Virtual Assistant) Do?” Users were also given the opportunity to expound on their response for each scenario if they wanted to. Demographic data was also collected and analyzed alongside participants’ responses to gauge for variations and trends. The results from this study will give policymakers and tech companies a peek into society’s ethical expectations as they relate to the moral decisions of VAPAs.
Furthermore, these findings will help illustrate a wide spectrum of concerns users have regarding the potential situational responsibilities of VAPAs as their pervasiveness continues to grow.
You can access the full research paper here.