Navigation auf uzh.ch
A new research initiative hosted by the Chair of Marketing for Social Impact investigates the adoption and effect of empathic AI.
Empathy, one of humanity’s defining traits, has become a focal point in artificial intelligence research. The newly launched AI Empathy Research Initiative, hosted by the Chair of Marketing for Social Impact at the University of Zurich, brings together senior researchers from Swiss universities to examine how AI can simulate empathic responses to enrich human-AI speech interactions. Led by Dr. Alex Mari (UZH), Dr. Ertugrul Uysal (ETH), and Dr. Fotis Efthymiou (HSG), the Initiative investigates how and when empathic AI—particularly voice assistants powered by generative AI—can impact human decision-making and emotional well-being. This project, supported by the UZH Foundation, is conducted in collaboration with computational affective scientist Dr. Jeff Brooks (NYU) from Hume AI.
Why Empathy in AI?
Empathy-the ability to recognize and respond to others' cognitive needs and emotional states-is a complex, multi-layered construct, comprising cognitive empathy (recognizing others’ emotions), affective empathy (emotionally resonating with them), and compassionate empathy (motivated support). Replicating empathy in AI is challenging and requires meticulous programming and ethical considerations. Recent generative AI (GenAI) assistants, powered by large language models (LLMs) like OpenAI’s ChatGPT and Anthropic’s Claude, aim to make interactions feel more natural, emotionally attuned, and context-sensitive.
The AI Empathy Research Initiative seeks to understand the role of empathic AI in influencing human decisions and behavior. By carefully examining user interactions with AI, the Initiative addresses questions crucial to shaping the future of AI—questions not only about how empathic AI can enrich our digital lives but also about potential risks and ethical implications.
Empathic AI in Marketing and Society
AI that “gets” you is no longer a futuristic dream. Advancements in GenAI allow voice assistants to interpret emotional cues, tailoring their responses accordingly. The Initiative’s focus is on understanding the broader societal implications of this trend, particularly in the field of marketing. How can empathic AI assist companies in creating customer experiences that resonate on a personal level? And how might it shape the ethical use of AI, guiding individuals’ decisions in a balanced way?
Empathic AI offers managers insights into creating more human-centric tools, but it also alerts policymakers to the regulatory needs surrounding AI’s influence on personal choices. As empathic AI shapes consumer experiences and social behaviors, the Initiative’s research contributes vital knowledge for both practitioners and regulators.
The Technology Behind Empathic AI
Recent developments, such as Hume AI’s Empathic Voice Interface (EVI), illustrate the technological strides in creating emotionally attuned AI. EVI can detect user emotions by analyzing vocal cues like tone, pitch, and intonation, enabling AI to respond with greater relevance. By pairing these emotional insights with language models like ChatGPT, empathic AI becomes more adaptable and responsive.
As voice assistants become increasingly capable of mirroring human expressions and language, the question remains: Can a machine ever truly embody the essence of human empathy? For many researchers, empathy is inherently a human experience—one that AI can simulate but never truly feel.
The AI Empathy Research Initiative, by focusing on the ethics and effectiveness of empathy simulation, envisions that voice assistants exhibiting all essential dimensions of empathy can be perceived as authentic and genuine by their users, especially when prior emotional bonds with the assistant are formed.
Testing Empathy in Voice Assistants
The first project of the AI Empathy Research Initiative is focused on developing and establishing a comprehensive framework for measuring and manipulating empathy in GenAI-based voice assistants. It aims to lay the groundwork for understanding empathy as a multidimensional construct. To achieve this, the project will develop ways to precisely adjust these dimensions within AI interactions. This involves creating specific prompts and responses that allow researchers to observe how users respond to different levels and types of AI empathy. The goal is to accurately capture how each facet of empathy influences decision-making in everyday scenarios, such as consumer choices or advice-seeking. In doing so, the research project seeks to shed light on empathic AI's practical applications and boundaries.
Shaping a Responsible AI Future
The AI Empathy Research Initiative is about more than making voice assistants seem relatable. It’s about envisioning a future where AI aligns with human values, acting as a supportive ally in decision-making and contributing positively to users’ lives. While empathic AI opens new possibilities for engagement, it raises ethical concerns. Could an empathic voice assistant influence users’ decisions in ways they might not intend, or could it empower them by offering truly considerate support? The Initiative explores these ethical dimensions, working to understand how AI might balance persuasive power with transparency and user autonomy.
Among its research projects, the team is set to launch investigations such as the following:
1. Encouraging Sustainable Consumption through Empathic AI: This project studies how empathic AI shopping assistants might reduce impulse purchases and promote sustainable consumption. It aims to identify the specific empathic interventions in AI interactions that are most effective at guiding consumers toward more mindful and reflective purchasing decisions. PI: Dr. Alex Mari.
2. Empathic Voice Assistants on the Frontlines: This project explores the role of empathy in the functioning and effectiveness of voice assistants in customer service and healthcare. We particularly focus on identifying the contexts in which AI empathy might be most beneficial as well as boundary conditions. This project aims to provide actionable insights into how to design and optimize empathic voice assistants for customer service and healthcare. PI: Dr. Ertugrul Uysal
3. Empathic Communication Styles as a Means to Mitigate Aggressive Behaviors: This project explores the phenomenon of online disinhibition, characterized by lowered behavioral constraints in digital environments, which can lead to aggressive and toxic behaviors. It aims to uncover successful empathy-based interventions to decrease aggression during speech interactions with AI and improve user experience. PI: Dr. Fotis Efthymiou.
Collaboration Opportunities
The team welcomes inquiries from senior researchers interested in collaboration on empathy-focused projects, technological partners developing software with empathic features, organizations implementing real-life empathic AI solutions for consumers, and institutions considering funding opportunities for these research activities.