Study abstract: Voice Similarity and its Impact on Cognitive and Affective Trust in Automated Vehicles
As self-driving cars move closer to everyday reality, one big question remains: Will people actually trust them?
A new University of Michigan study finds that the answer may lie in the sound of the car's voice.
When the voice of an automated vehicle matched the driver's own-particularly their gender-people reported higher levels of trust. Gender similarity strongly influenced both logical confidence in the car's abilities and emotional connection, while age similarity mainly shaped the emotional bond.
The findings suggest that something as simple as voice design could play a powerful role in making AVs feel not just smart, but relatable.
Research shows that widespread adoption of AVs faces significant hurdles, mainly stemming from public skepticism about their safety and operational reliability. Voice gender is a debated element of voice design in AI technologies. Many voice assistants such as Siri and Alexa default to female voices, often drawing on anecdotal evidence suggesting that users across diverse cultures generally prefer female voices.
The current study, released at the Proceedings of the Human Factors and Ergonomics Society annual meeting, looked at how the voice of an AV affects two types of trust: cognitive trust (believing the car is smart and reliable) and affective trust (feeling an emotional bond with it).
Researchers used an online study with more than 300 U.S. drivers, who signed up online and learned how the self-driving car worked before starting. They watched six short videos of the car driving in different places, with the car's voice explaining what it was doing. After each video, they rated how much they trusted the car and then shared a bit about themselves.
When the car's voice matched both the user's gender and traditional gender roles, people trusted the car more, both logically (cognitive trust) and emotionally (affective trust).
But if the voice only matched gender and not gender role expectations, it mainly boosted emotional trust, not logical trust. The results show a challenge: Using gender roles may build trust but also risks reinforcing stereotypes, so developers should consider options like customizable voices, gender-neutral designs, or even nonhuman sounds.
"These findings highlight the importance of considering gender in AV voice design for conveying critical driving information and reveal how societal stereotypes shape AV design," said Lionel Robert Jr., U-M professor of information and robotics.
In short, the way a self-driving car "talks" really matters, researchers said.
"Designing AV voices to feel more personal and relatable could make people more comfortable trusting them," said the study's lead author Qiaoning (Carol) Zhang, assistant professor of human systems engineering at Arizona State University.
The research is also important for those not interested in buying AV cars.
"Even if you never plan to own a self-driving car, you will almost certainly share the road with them," said co-author X Jessie Yang, U-M associate professor of information and industrial and operations engineering. "Building trust in these vehicles is crucial for public safety and the seamless integration of automated cars into our transportation system."
