Have You Ever Said the F-Word to Alexa? : Why People Abuse AI

Alexa introduced a new function called ‘Politeness Feature’ that provides a feedback for users especially for kids, who ask Alexa with the word please. When a user adds it after their commands, Alexa responds by saying “Thanks for asking me nicely”. Furthermore, when you say “Thank you, Alexa”, she gives answers like “No worries” or “You’re welcome”.

The background story of this update interested me. A while ago, parents claimed that Alexa is spoiling their kids’ language behavior because she always gives kind feedback, even when kids talk to them rudely and aggressively. It makes so much sense that since having Alexa at home is similar to owning a personal assistant who always obeys you and never says “No”.

The problem is that this phenomenon is not limited to kids. Adults also use them, and the way they’re talking to Alexa is not very polite. Some users even swear to assistants. You know, we sometimes lose our tension when machines don’t understand our orders.

To be fair, using bad language is a relatively benign way of bullying a virtual assistant. Some are even going through sexual harassment. A research indicates that 5% of all chat logs between GPS voice and truck drivers was a sexual abusing, and the number may go higher when it comes to all virtual assistants.

Then, Why People Talk Aggressively to AI?

There are two main reasons why people abuse non-human agents. The first reason is Catharsis hypothesis. People use agents as a punching bag to release their anger induced from other sources. Honestly, AI agents can be a perfect object for venting our anger. The fact that they don’t have a soul releases users from the feeling of guiltiness when they bully non-human agents.

Secondly, there is a frustration hypothesis. It says that users act rudely because machines don’t fulfill their expectation. You can imagine the situation of someone swearing to a damaged computer. If Alexa doesn’t get what users said, they can blame her for it.

These two exlanations above are both persuasive. However, I’d like to add a new hypothesis, known as the empowerment hypothesis. Since most AI agents have a submissive and kind personality, the power dynamics between users and AI is formed as a master-and-slave relationship. It gives user a full authority and the right to do whatever they want. So, the tougher they treat assistant, the more power they can feel during the interaction.

The very thought of interacting with virtual assistants is an entirely new concept, and so far, there’s no consensus of how to talk properly to them. However, some researchers have pointed out that the way people treat non-human beings can influence their real-world interactions. You know, everyone’s true colors show eventually! Maybe now is the time to consider the side effects of having a humanlike, kind, and alway-smiling personal assistant.