Agents and Culture: Competition, co-existence or reconciliation?
Vienna Workshop "Agent Culture", 24-25 August 2001
Kerstin Dautenhahn
University of Hertfordshire
One might question whether the future will look like many agentresearchers tend to paint it, namely a future with numerous anddifferent types of robotic and software agents populatinghuman-inhabited real and virtual spaces. However, asuming that thismight happen at some point in the future, one might ask: is it afuture that is desirable, can we and do we need to make an effort inshaping it?
First of all, aren't agents just another pience of technology andinvention that will certainly change our lives but that we are happyto accept (such as telephone, trains, cars etc.)? Is there anythingspecial about agents? The argument I like to present is saying thatindeed, this new generation of agents that are emerging is special.These agents are different from cars and trains and typewriters sincemany of them have the primary, if not sole purpose of beingpersuasive machines, machines that change our beliefs, attitudes andbehaviour, agents that humans relate to and establish a"relationship" with. Pet robots and embodied conversational agents,just to give a few examples, embody representations of what weconsider "social" or "appropriate" behaviour, they convey meaning andvalues. Such agents are cultural artifacts, and they play a role inchanging human society and culture as an active member of thissociety: they are social actors, and it is because of this that theyare different from tools or artworks that influence implicitlythrough usage and interpretation.
An important danger of a future where most of our dailyinteractions might be with and mediated by artificial agents, is theloss of cultural diversity. Children in Tokyo watch the same cartoonmovie and play with the same Aibo robot as children in Los Angeles,Sydney and London (note, that we use the "Aibo" as a metaphor forrobot pets, in fact most owners of Aibos are adults). Advantages ofsuch a development are that children can in this way communicateabout shared experiences. Disadvantages are that the diversity oftraditional toys and patterns of play and interaction might get lost,being replaced by stereotypes that are being understood easily, basedon compromises. Biological evolution depends on the generation ofdiversity, the fact that offspring no matter how similar they are totheir parents are never identical. Diverse populations are naturespool for new "ideas", new designs for body plans and adaptations.Similarly, cultural evolution depends on diversity, too. Aconvergence of cultures and a mass extinction of local diversities isalready under way, cf. the spread of "Starbucks" or "McDonalds"across the globe. If children everywhere in the world (or at least inmany countries) use the same toys, learn with the same educationalsoftware, play the same games, where should new ideas come? "Novelty"is often not that novel, is often a result of blending concepts andideas from different domains. I will argue that similar to the massextension of biological evolution that is under way and greatly dueto human interference, a mass extinction of cultural diversity hasalready begun, and agent technology might contribute to itsaccelaration.
How can such tendencies be counterbalanced: Firstly, agents needto be explictly perceived as culturally situated artifacts, asartifacts that convey meaning and values. Secondly, agent design assuch needs to consider not only culturally adaptive agents, butagents that support cultural diversity, that embody meaning andvalues differently in different communities. Agent systems have beenused for people to express their values and facilitate communication,and establish a shared understanding, but it seems equally importantfor agent systems to support values and meaning that is specific toparticular communities, supporting a diversity of understanding andidentity rather than a shared understanding based on levelling outdifferences. Communication and shared understanding across culturesis important, as long as these cultures still exist. Agents couldplay an important role in supporting cultural diversity. To give anexample: a mechanic toy produced for the mass market cannot adapt tolocal values and meaning. A computational or robotic agent that canlearn and adapt its behaviour and appearance, and that behavesautonomously as a persuasive machine can embody values and meaning ofparticular communities and cultures. Different from the mechanicaltoy it might actively contribute to the conservation of culturaldiversity.