Ruling the Future: Implicit Motivators of People's Gendered Perceptions and Preferences of Voice Assistants

Yükseköğretim Kurumları Destekli Proje, 2024 - 2025

The gendering of artificial intelligence (AI) technology has been increasingly a part of scientific interrogation and information

retrieval, personalized suggestions, and daily decision-making. The most available example of this is the use of voice assistants

(VAs). The most renowned VA tools such as Apple’s Siri, Amazon’s Alexa, or Microsoft’s Cortana are female by default.

However, assigning female voices to VAs carries risks for maintaining gender stereotypes such as women are submissive, obey

commands, care for others, help those in need, etc. Research also shows that people prefer female VAs over male VAs, and see

them as more trustable and helpful. Stereotypes’ implicit effects may maintain traditional gender stereotypes and result in

negative effects on gender equality in the long run. Progressive research in science and technology has started exploring

whether using gender-neutral voices offers better practices for equal societies. However, most AI-related research has been

conducted in Western-based, English-speaking cultures. This risks jeopardizing the possible use of gender-neutral or even

gendered VAs in different languages as the listeners' perceptions, tone, voice pitches, or phonetics may not guarantee how VAs

are understood. In light of these debates, this research aims to understand people’s perceptions of VAs and the motivations

behind their gendered preferences by using the psychology field's knowledge and methodology. For this aim, we propose two

experimental studies to be conducted in two languages: Turkish and English.

In study 1, we aim to extensively explore potential reasons behind people’s perceptions and preferences of gendered (female

and male) and non-gendered (gender-neutral) voice assistants among Turkish participants. We explore the effect of people’s

gender stereotype endorsements, the gender identity (female, male, or gender-neutral) of the VA, and the stereotypical message

content (feminine, masculine, or neutral) given by the VA with between-subjects experimental design. We will collect data from

250 participants in a controlled and computerized lab environment. Study 1 will serve as a baseline study for Study 2, which aims

to compare the additional role of VAs’ language among English-speaking participants. Study 2 will also serve as a replication of

the first study in different cultures and languages. The data will be collected from 250 participants enrolled in the Prolific data

collection tool, which is an online professional pool for representative participants in England. All the materials and the formal

(not physical) procedures will be the same, allowing the project to compare and replicate the findings in two cultures. By doing

so, it has the potential to unveil a new research area to explore in future studies. Both studies are exploratory in nature as

scientific findings on VA gender preferences have been inconsistent. The findings will pave the way for exploring the choice

patterns, especially for gender-neutral VAs to lead more bias-free production of knowledge in future research.