Opportunities and Challenges Facing AI Voice-Based Assistants: Consumer Perceptions and Technology Realities: An Abstract

Abstract

Where AI has become especially pivotal for users’ interactions is in the case of voice-based assistants (VAs), such as Apple’s Siri and Amazon’s Alexa. From what is initially perceived as being a tool to play music, read out news reports and set timers, VAs have developed considerably in recent years and their functionality goes way beyond initial perceptions. For example, Amazon Alexa will be able to give out health advice to users in the United Kingdom as health-related questions will be automatically searched for using the official NHS website (MIT Technology Review, 2020). Some reports go as far to discuss how Amazon has plans to be able to run someone’s entire life from the Alexa on the basis that the systems are getting so sophisticated and the data being collected is so vast that the Alexa will be capable of predicting needs (Hao, 2019). Nevertheless, users still appear to be resistant to use VAs, with some even reluctant to engage with this technology entirely (PwC, 2018). Most of these constraints to VA adoption primarily relate to lack of trust, perceived data privacy and security concerns and lack of knowledge or understanding.

Despite the benefits and opportunities of AI software, it is inherently limited by capabilities surrounding planning, reasoning, knowledge, natural language processing, ability to move and to empathise. It is this lack of emotional connection that is a fundamental component of users being less trusting towards AI voice-based assistants. This research investigates the role of emerging capabilities, privacy concerns and trust towards VAs using a mixed-methodology design. Results from the qualitative element of the study reveal that it is more about trusting the capabilities and functionalities than being scared over personal data. This theme will subsequently be examined further in a follow-up experimental study to examine the role of context towards these perceptions. The focus of the experimental design will subsequently consider the direct relationship between perceived behavioural control and intention to continue to use VAs with consideration into the indirect effects of privacy concerns and trust and the moderating role of perceived capabilities.

The findings, thus far, show how VAs, as new technologies and leaders of machine learning capabilities, can enhance customer experiences, improve customer relationships and add value to firms. As privacy concerns are often dismissed, if customers benefit from the sharing of their information, yet their capabilities being under-appreciated, firms can focus attention to the high-quality functions of these devices to encourage their more frequent use.