Why Make Bots Believable?

A little story from Aladdin’s magic carpet.

Do we all remember the good old Disney movie Aladdin? Back then, Disney showed us how believability can be translated into characters of non-human nature, with just a magic touch of human personality. If you remember Aladdin’s magic carpet you already know what I am talking about. A quote by Loyall in 1967 draws a comparison of virtual agents and the magic of personality: “Believable agents are personality-rich autonomous agents with the powerful properties of characters from the arts” (…) It is shown even more strongly by the character of the Flying Carpet in the Disney animated film Aladdin. It has no way of being realistic, it is a totally fantastic creature. In addition, it does not have many of the normal avenues of expression: it has no eyes, limbs nor even a head. It is only a carpet that can move. And yet, it has a definite personality with its own goals, motivations and motions.” (cit.)


If we think about bots, many challenges are driven around the notion of making bots truly intelligent. But what what happens if we can make them, alongside, more human-like? And this is what struck me and inspired me to dig deeper into the literature of HCI, specifically in the context of conversational agents. Merriam-Webster defines believability as capable of being believed especially as within the range of known possibility or probability.

This short text will fast forward explain why it is good to build believable bots.

Trust. In Human-Computer Interaction (HCI) studies have touched upon the notion of credibility (i.e. believability) in terms of computing products. In 1999, Fogg and Tseng have discovered that believability in general tends to be described as a perceived quality and is defined with notions such as trustworthiness and expertise.

Value. When it comes to human-bot interaction, believability can be considered through different characteristics, such as personality. Hingston in 2012 stated that bots bring more value if they operate at a human level. For example, if a bot possesses an extroverted personality, it can be referred to as the power, status, or control factor, and it ranges from dominance to submissiveness.

Bonding. As social beings, humans have the tendency to group things with common characteristics. One of these aspects is attachment or binding, and it draws from the psychology of interaction among humans, which describes the dynamics of long-term and short-term interpersonal relationships. So if the bots and humans have the same view of the world, we could build towards building lasting interpersonal relationships.

Higher agency. Anthropomorphism is the tendency of individuals to believe that computers are like people. A study by King and Ohya in 1996 compared the perception of human and anthropomorphic representations. Even though human forms are more appraised in icons, all of the anthropomorphic representations were judged to be more intelligent and capable of higher agency than the other stimuli presented.

Richer experiences. Also, in games, bots are rewarding in richer and more engaging experiences to players, therefore Hingston underlines the importance of believable bot behaviour by encoding psychosocial elements (such as aforementioned personality, social role awareness, emotions, etc.)

So let’s all dig into that magic that Disney showed us some years ago. If we think about it, Aladdin’s carpet seems to be very intelligent and even though it cannot speak, the carpet presents itself as very shy at the beginning, but then very loyal, romantic, funny, cheerful, and heroic (saving Aladdins life a few times here and there). So it seems that if something behaves in a human manner, we treat it as human and for that same reason, we should start building bots that are believable. Aside from making them truly intelligent and capable of understanding intents, we need bots that present personality, express emotions, and are capable of building relationships with humans. To conclude, this is why we are to design believable bots:

Believability can be perceived as trustworthy and as an expert quality.

Believable bots bring more value to humans if they operate at a human level, for example, by projecting personality.

As humans are social beings, we tend to bond with things that behave similarly to us.

As anthropomorphic agents, believable bots can be judged to be more intelligent and capable of higher agency compared to human agents.

In games, believable bots bring richer and more engaging experiences.

Thank you for reading,

Let’s talk more on Twitter (@iStefan08) and LinkedIn (/stefanmanojlovic)

Written by:

Stefan Manojlovic, UX Researcher at Bellabeat