Repository logo
  • English
  • Deutsch
Log In
or
  1. Home
  2. HSG CRIS
  3. HSG Publications
  4. Designing Relatable AI: A Theory of Mind Perspective of How Conversational AI Shapes Technology-Mediated Consumer Behavior
 
  • Details

Designing Relatable AI: A Theory of Mind Perspective of How Conversational AI Shapes Technology-Mediated Consumer Behavior

Type
doctoral thesis
Date Issued
2022-02-21
Author(s)
Bergner, Anouk Samantha  
Abstract
As technology is evolving to appear progressively more human-like, next generation conversational artificial intelligence (AI) is profoundly impacting human-technology relationships. Conversational interfaces (CIs) effectively mimic human conversation in either written (as with chatbots) or spoken form (as with digital voice assistants), taking over a broad range of tasks from acting as personal assistants, to selling products to consumers, or recommending investment strategies in the financial industry. Despite this fascinating and broad range of applications, one unifying element across the evolving conversational AI landscape is the ability to interact with consumers in what was previously considered the hallmark of humanity: language and dialogue. This raises foundational questions of how consumers perceive progressively more humanized AI, as well as the behavioral consequences for consumers. This dissertation develops a novel theoretical framework of human-technology relationships in the context of conversational AI, building on and integrating fundamental research on the theory of mind in psychology, foundational principles of human dialogue in linguistics, and emerging literature on technology-mediated consumer behavior. The framework and the underlying hypotheses are developed and tested in four independent paper projects, across a broad range of methodological approaches (deep contextual language models and experimental research designs), study settings (both in the field and the lab), and participant samples (from financial investors to online shoppers). Together, these papers demonstrate how interacting with conversational AI fundamentally shapes consumer decision-making and assess the underlying psychological mechanisms and moderating factors influencing the formation of intimacy and trust with such interfaces (and the brands or firms they represent). Furthermore, they provide novel methodological avenues, compared to traditional consumer research, to unobtrusively identify variations in the attribution of a mind to these interfaces. The overarching objective of this dissertation is to contribute to an emerging and interdisciplinary field of research on how humans relate to and are influenced by increasingly humanized AI in a digitally transforming society.
Abstract (De)
Die rasche Entwicklung von Technologien wie künstlicher Intelligenz (KI) auf immer menschlichere Art zu kommunizieren, hat tiefgreifenden Einfluss auf die Beziehung zwischen Mensch und Technologie. Dialog-basierte KI in Form von Conversational Interfaces (CIs) ahmen menschliche Konversationen in schriftlicher (wie bei Chatbots) oder gesprochener Form (wie bei digitalen Sprachassistenten) nach und übernehmen ein breites Spektrum an Aufgaben: von der Funktion als persönlicher Assistent über den Verkauf von Produkten bis hin zur Empfehlung von Anlagestrategien in der Finanzbranche. Obwohl der Anwendungsbereich dialog-basierter KI sehr breit ist, gibt es ein übergreifendes Element, welches diese Anwendungen vereinigt: ihre nahezu menschliche Fähigkeit, mit Konsumenten in natürlicher Sprache zu interagieren. Dies wirft grundlegende Fragen darüber auf, wie diese zunehmend menschlichere KI wahrgenommen wird und welche Konsequenzen dies für das Verhalten der Konsumenten hat. Die vorliegende Dissertation entwickelt ein neues theoretisches Fundament zur Erforschung der Beziehung zwischen Mensch und KI im Kontext der CI basierend auf Literatur zur psychologischen Theory of Mind, Grundprinzipien menschlichen Dialogs in der Linguistik und der Forschung des technologie-getriebenen Konsumentenverhalten. Das theoretische Fundament und die daraus abgeleiteten Hypothesen werden in vier unabhängigen Forschungsprojekten entwickelt und mithilfe mehrerer methodischen Ansätze (von Deep Learning Sprachmodellen bis hin zu experimentellen Forschungsdesigns), Studiensettings (sowohl Feld- als auch Laborstudien) und Stichproben (von Finanzinvestoren bis hin zu Online-Käufern) getestet. Zusammen zeigen diese Arbeiten wie die Interaktion mit dialog-basierter KI die Entscheidungsfindung von Konsumenten grundlegend beeinflusst. Sie erforschen die zugrundeliegenden psychologischen Mechanismen und moderierende Faktoren, welche die Bildung von Intimität und Vertrauen mit dialog-basierter KI (und den Marken oder Unternehmen die sie repräsentieren) beeinflussen. Sie bieten im Vergleich zur traditionellen Konsumentenforschung neuartige Methoden, um die Zuschreibung eines Menschlichen Verstandes zu dialog-basierter KI valide zu identifizieren. Das übergreifende Ziel dieser Dissertation ist es, einen Beitrag zu einem sich neu entwickelnden und interdisziplinären Forschungsfeld zu leisten, das sich mit der Frage beschäftigt, wie Menschen in einer sich digital transformierenden Gesellschaft mit zunehmend menschlicher KI umgehen und von dieser beeinflusst werden.
Language
English
Keywords
Chatbot
Künstliche Intelligenz
Automatische Sprachanalyse
Verbraucherverhalten
Kaufentscheidung
EDIS-5163
theory of mind
brand intimacy
machine intelligence
natural language processing
deep learning
Natürliche Sprachsysteme
consumer-brand relationships
digital voice assistants
interpretable machine learning
technology-mediated choice
mind perception
language models
investment automation
conversational interfaces
smart objects
consumer financial decision making
trust
robo advisors
Chatbots
HSG Classification
not classified
HSG Profile Area
None
Publisher
Universität St. Gallen
Publisher place
St.Gallen
URL
https://www.alexandria.unisg.ch/handle/20.500.14171/108948
Subject(s)

social sciences

Division(s)

IBT - Institute of Be...

Eprints ID
265902
File(s)
Loading...
Thumbnail Image

open.access

Name

Dis5163.pdf

Size

7.88 MB

Format

Adobe PDF

Checksum (MD5)

52627618341769df808c285390a4665e

here you can find instructions and news.

Built with DSpace-CRIS software - Extension maintained and optimized by 4Science

  • Privacy policy
  • End User Agreement
  • Send Feedback