Jessica Barfield and robot Misty
As once unreachable technologies like artificial intelligence become commonplace, many are considering how these can be implemented in their everyday lives, and some are left wondering whether they can even be trusted.
Jessica Barfield’s (assistant professor, SIS) research focuses on human-computer interaction (HCI) and human-robot interaction (HRI), which refer to human interaction between software and physical, autonomous robots and AI entities respectively.
Barfield earned her doctoral degree in information sciences at the University of Tennessee-Knoxville, where she first began researching HCI and HRI. In her research, Barfield observes biases people show toward technology based on its perceived social characteristics. Her interest in the field was sparked when she noticed how users select and interact with different nationalities and genders of Apple’s virtual assistant, Siri.
“People don’t always think about the humanness of human-robot and -computer interaction,” Barfield said. “I want to make sure that the designers of these technologies are aware of biases and keep in mind who their diverse user audience is.”
Being faculty, Barfield has access to myriad resources that allow her to examine biases in HCI and HRI. Quantitative surveys administered virtually by Barfield have presented participants with audio files, photos and clips of robots, posing questions like “How does this make you feel?” and “Do you trust it?” And with hands-on tests, Barfield can bring in actual robotic hardware for participant interaction.
Enter Misty, the small-but-mighty programmable robot Barfield shares with her alma mater, UT’s School of Information Sciences. Equipped with depth cameras, touch sensors and interchangeable limbs, Misty has the capabilities to move on preset paths and react to participants with customizable eye movements.
Additionally, Misty can speak and understand a host of languages and can even speak with an accent. Barfield said chatbots like OpenAI’s ChatGPT can be installed on Misty to accommodate a more back-and-forth conversation with participants.
Barfield found that people treat technology like they do actual people, noting real-life gender and race biases reciprocated toward software and robots. “Somebody could be interacting with a female-voiced chatbot or technology like Misty and have gender biases that we typically attribute toward females,” Barfield said. “But if I set [Misty’s] voice to a deeper tone and call her ‘Robert,’ people sometimes react better, and trust Robert more.”
Looking forward, Barfield not only thinks these technologies will become more customizable but also that people will start to trust them more.
“Right now, [industry] focus is on creating tech that is cool and can fulfill needs,” Barfield said. “But I think the missing part is the human connection. It’s important, especially when tech is being created for more vulnerable populations, like people with disabilities, that designers are creating things people trust and want to use.