Humanlike AI assistants may soften security concerns through enjoyment, study finds

Making an AI assistant feel more human may change how users weigh its security risks—not directly, but through how much they enjoy it. A study of 277 voice-assistant users links greater enjoyment to lower perceived security risk and less resolve to stop using the product, according to an October 29 University of Kansas account.

The research, published in Information & Management, examines anthropomorphism: giving a nonhuman system humanlike qualities, such as a friendly voice, appearance or personality. Co-author Andrew Chen, a professor at the University of Kansas School of Business, describes a familiar social response carrying over into interactions with machines: friendliness can encourage trust.

Following the path from enjoyment to risk

The team analyzed survey responses using structural equation modeling, a statistical method for examining relationships among measured factors. It supplemented that analysis with robustness checks and follow-up analyses. The evidence concerns existing voice-assistant users and their reported attitudes, rather than a security audit of the products they used.

Its central distinction is between a direct relationship and an indirect one. Humanlike design significantly increased perceived enjoyment. That enjoyment, in turn, was associated with reduced perceived security risk and weaker intentions to stop using the assistant. Humanlike qualities did not directly reduce perceived risk, Chen said.

Those outcomes describe how people felt and what they intended to do. Lower perceived risk does not mean the technology became safer, and less resolve to quit is not the same as observing whether someone actually kept using it. The finding is about the relationship between product presentation, enjoyment and judgment.

It’s very difficult for the user to assess what’s behind the technology or system. That’s a disadvantage. But if you talk to people or something humanlike, you probably feel more comfortable.

Andrew Chen, University of Kansas School of Business

When feelings become a shortcut

To explain the mechanism, the researchers examined emotional and cognitive appraisal: how people feel about an assistant and how they evaluate it. Chen said earlier research typically treats those as parallel paths. His team instead found that the emotional response interacts with the more deliberative assessment.

In Chen’s explanation, feeling comfortable with or attached to an assistant can become a mental shortcut for judging the technology behind it. Users then give less attention to evaluating it. The team characterized the relationship as quadratic rather than a straight line; Chen described risk perceptions falling more as comfort increased.

A design recommendation with a tension

Chen sees practical value in making assistants more enjoyable. He urged designers to look beyond function and devote more effort to the pleasurable side of interaction, while stressing that an assistant still has to work. His recommendation is not to replace useful features with personality, but to give comfort and enjoyment more design attention.

He also described convenience as voice assistants’ main benefit: people can get answers and suggestions without going to a computer. Yet his own use does not erase other people’s concerns. Chen recalled a visitor asking him to unplug an Amazon Echo because the visitor worried about it recording things.

The article, titled Sounds Good and Feels Better — How Anthropomorphism Influences Artificial Intelligence Voice Assistant Use, therefore puts two aspects of design together: making interaction more comfortable and understanding how that comfort affects scrutiny. Its reported result ties enjoyment not just to a more pleasant experience, but also to users’ willingness to continue.