Design challenges for sustainable interaction with social artificial agents

With increased digitisation and automation also comes increased interaction with artificial agents, such as industrial robots, voice assistants, chatbots and household robots. In order for these to develop and create value, we need to examine what is required to develop artificial agents so that they meet the requirements for good interaction, and what can stand in the way of the agent being able to be used in the long term. A controversial aspect is how human-like social artificial agents or robots should be and how this affects the interaction and reliability towards the agent.

This article is based on a bachelors thesis in the field, and the article highlights aspects such as human-like design, social presence and expectations, as well as how these affect how the interaction with social artificial agents is experienced by people. The work showed that expectations of the interaction with the artificial social agent affected the experience. The experience was also linked to the agent's social presence and the discrepancy between the agent's human appearance and his behavior.

Woman 163426 1920

Artificial agents in social environment and working life

Artificial agents have gone from being a fictitious idea to starting to be implemented in several areas. Despite this development, we still have limited knowledge of how these agents should be designed to create sustainable interactions with humans. Social artificial agents can be the next commonly used interface because it is adapted to the human social environment, which means that the agent can be used in both working life and in people's homes. For this to be possible, however, we need to understand what affects our ability to interact with these agents, and how the agents' designs affect us. In Japan, for example, the development of social robots has progressed rapidly and become an important part of the social environment. Therefore, the issues are important for our future social environment, but also in the industry and working life of the future.

The Uncanny Valley and human-like artificial agents

Whether social artificial agents should have a human-like appearance is debatable as these agents can create feelings of discomfort in the human user. This phenomenon is called The Uncanny Valley. The Uncanny Valley is a well-known design principle based on the idea that realistic, human-like design that is at the same time distinguishable from its human counterpart can arouse strong feelings of discomfort and eerieness in the user. Therefore, human-like design may have the opposite effect than intended, that is, it may complicate the interaction between human and artificial agent. Another disadvantage of human-like agents is that an incorrect image of the agent's capacity may arise due to the expectations that arise in the user. This is because people carry knowledge from others, similar situations, and because people have difficulty seeing individual abilities in isolation. For example, if a social agent can smile like a human, or has the ability to recognize faces, it can lead to expectations of emotional intelligence or complex interaction characteristics as we are used to this in connection with human interaction situations. In addition, if something in the agent's design indicates advanced abilities, it is also expected to have more basic properties, which can create problems in interpreting the agent's ability and difficulty in interacting with it. However, humane design can have its benefits. One of the advantages of this type of design is that it can improve the interaction with people as the agent gives a familiar impression and a sense of social presence. Human design can also facilitate the understanding of the agent's functionality and area of ​​use as the design is based on human characteristics that we recognize from interaction with other people.


To create sustainable interactions with social artificial agents, we need to understand how the design can be adapted so that it creates the right expectations in relation to the area of ​​use and the intended interaction. Therefore, we need to think about in which situations a human-like agent is preferable to other alternatives. The artificial agents of the future need to be trustworthy and reliable, and must meet the needs defined by those who will interact with them. The interaction experience with human-like agents seems to be more multifaceted than previous research has established, and thus we need to expand the knowledge about which emotional reactions can be limiting for the interaction, and what these reactions are based on. Social artificial agents are still a new kind of interface that has not yet been integrated into society. Therefore, it is unusual for users to have experience of continuous interaction with these interfaces. Because of this, we need to consider an additional factor that may affect our ability to interact with social artificial agents - our ability to adapt and get used to new things. When the pleasure or discomfort of the news is not left, what does the experience of the interaction look like then? This will also be crucial for the sustainable interactions of the future.

Curious about more?

You can read about the social robot Furhat, The Uncanny Valley, and the article Conceptualizing Embodied Automation to Increase Transfer of Tacit knowledge in the Learning Factory.

Link to the bachelor thesis performed by Anna Johansson och Elin Örnberg.

Anna E Lin
Anna Johansson and Elin Örnberg


This article is tagged with these tags. Click a tag to see all the articles with this tag.

This article is categorised as Intermediate  |  Published 2021-09-22  |  Authored by Sandra Mattsson