A Cognitive and Socio-cultural Perspective on the Tendency to use Gmail’s Smart Reply-like AI-based Texting Features

Abstract

Introduced by Google in the year 2016, Gmail Smart Reply gives reply suggestions to users through deep neural networks, based on its ML model trained on a massive database collected earlier. By 2017, Smart Reply was already sending about 6.7 billion email replies on behalf of humans. The present study explores the tendency of email service users from India to use features like Gmail’s Smart Reply. The final data comprised 167 participants, including 92 students and 75 working professionals from India. The simulation section of the study posed six hypothetical Emails to the respondents. Students and professionals received email contexts that were relevant to their experiences respectively. To account for the message’s nature, an equal number of formally and informally worded emails were used. The direction of communication was incorporated by making participants respond to Emails from a subordinate, a peer or colleague, and a senior. Gmail Smart Reply-like short responses were provided and participants had the option of choosing these or typing their own responses. Findings showed that over 80% of participants used the Smart-Reply-like option rather than typing their own replies. The cognitive and socio-cultural explanations for this tendency are explored. The study has socio-cultural implications in the context of democratizing AI-based language solutions and ensuring the sensitivity of the solutions to regional needs. While features like Smart Reply provide supportive ecosystems for non-native English speakers, its consequent domination over naturally produced language may have implications for computer-mediated communication and cognitive linguistic studies in the future.



Author Information
Sairaj Patki, FLAME University, India
Omkar Joshi, FLAME University, India
Ritwika Das, FLAME University, India

Paper Information
Conference: ACP2023
Stream: Linguistics, Language & Psychology/Behavioral Science

The full paper is not available for this title


Virtual Presentation


Comments & Feedback

Place a comment using your LinkedIn profile

Comments

Share on activity feed

Powered by WP LinkPress

Share this Research

Posted by James Alexander Gordon