The aspiration of combining an anthropomorphic form and a human-like performance on a machine is lately reaching new levels, with the wide adoption of virtual assistants. The former, going beyond the notion of purposeful, assisting duties are sometimes named anew “digital humans”, whetting the appetite for experimentation beyond the horizons already calculated for them.
Our artistic experiment explores whether the implementation of two discreet and specific theatre directing theories, such as Stanislavski’s and Artaud’s, holds interest in the organization of the dramatic and performative behaviour of digital humans and the production of differentiated, aesthetically coherent outcomes. It also suggests the exploration of new dramatic/performative perspectives for IVAs.
Τhe UneeQ Digital Humans platform was chosen as the main tool for the experiment after some research as it meets the basic requirements, i.e. real-time human-computer interaction with natural language (spoken & written), machine learning, variety of emotional responses and realistic graphics. For the current research stage, a limited free trial edition of the UneeQ platform was deemed sufficient. Additionally, Google Dialogueflow was used for the design of the dialogues.
Two different perspectives in the design and implementation of a prototype of a virtually iconic embodied chatbot (digital human) have been followed, Stanislavski’s realistic approach and Artaud’s surrealistic one, with evident impact both on a dramatic (script) level as well as on the performative, behavioural level of the digital human, manifested with the attribution of “emotes”, leading to two different interactive experiences.
To create an interactive drama, two scripts, inspired by the methods of the two theatre directors, were conceived and transformed into multi-linear, following a logic of “multiple improvisational rehearsing”, while using the aforementioned technologies. For the purpose of the experiment, the two scripts aspired at delivering the same meaning load in both versions, following different dramaturgical aesthetics in each case. A life-like scenario of interacting with a bar assistant was chosen for both versions. The central scenario wants Claire, the character of our digital human, to be an assistant at the fictional Kouros lounge bar on the island of Paros. The name of the character was kept the same and some information retained intact, such as opening hours and music. The strategy of changing solely the character of the text and the emotes of the digital human was preferred while the spectacular aspect was kept identical (space, digital human face, hairstyle e.t.c). The emotional aspect of human-computer interaction was addressed by attributing to the digital human emotes from the range of those offered by UneeQ, namely “happy”, “excited”, “joking”, “empathetic”, “agreeing”, “enquiring”. In the Stanislavskian approach, the attribution of emotes mostly followed the text naturally, whereas in Artaud’s case this attribution was more arbitrary. In the case of the simulation of the expression of negative feelings (e.g. anger, indifference e.t.c.), because of the restrictions applying on the available emotes, new creative combinations had to be tried out.
As the aforementioned technologies were used on trial editions and limitations were imposed, the demo consists of videos that document the two designed as differentiated interactive experiences and not the interactive experiences.
UneeQ Digital Humans: https://digitalhumans.com/
Google Dialogflow: https://cloud.google.com/dialogflow