Researchers Explore How Personality and Integrity Shape Trust in AI Technology

AI has become an important part of our lives and we cannot escape it no matter how hard we try. But the question arises if people trust it enough and if they do, what influences them to make this decision? Researchers from University of Basel conducted a study to find out to what extent do people trust AI chatbots and what factors does it depend upon. For the study, the researchers made up a text based AI platform called Conversea, and analyzed the interactions between the chatbot and the users.

There are a lot of factors that make us trust something. It can be our own personality, how others behave with us, others’ personality and also some specific situation that calls for us to trust someone. The ability of people to trust someone develops from childhood and helps us decide how open we want to be with someone. The researchers say that the factors which play a role in trusting someone also play the same role in our trust in AI systems.

The characteristics most important for trust are integrity and competence and these two help humans evaluate if an AI is reliable or not. The study also found that participants do not think of AI in the light of the company which created it, they think of it as a whole separate unit. Impersonal and personalized chatbots also play a role in our perception of trust in them. If a chatbot is referring to us by our name and also mentions previous conversations, the participants say that the AI chatbot is competent and kind.

When an AI chatbot is personalized, the users think of it as a human and that's why they tend to share more personal information with it and want to use it more. But the study found that there was no difference in trust in personalized and impersonal chatbots by the participants. The study says that for trust to develop, integrity is the most important factor so developing integrity in AI chatbots should be prioritized. Most of the lonely people have started relying on AI because they seem personalized to them.

The researchers of the study said that they think that AI systems should be reliable above anything else. The researchers haven't said anything about whether trusting AI is good or bad, but they think that too much usage of AI as a friend can isolate us from our social environments. AI chatbots should always give advice to users along with the consequences and risks of it. They should also stop inventing answers and just tell the users that they don't have an answer for their question to give them some reality check.

Image: DIW-Aigen

Read next: China’s AI Chatbot Market Sees ByteDance’s Doubao Leading Through Innovation and Accessibility
by Arooj Ahmed via Digital Information World