Information Technology Reference
In-Depth Information
In “Writing and representation,” Agre examines the stories AI researchers
tell about human representation use, and counters with alternative stories
about representation use.
In “Schizophrenia and narrative in artificial agents,” Sengers critically anal-
yses the way in which AI researchers talk about their agents, discovering simi-
larities to descriptions of schizophrenic patients. These stories reveal the extent
to which AI researchers think of and build their agents as simple mechanisms,
although they should appear as complex, living beings. She suggests a compro-
mise approach, in which the agent is thought of as a mechanism which can take
advantage of the human propensity to create narrative explanation to create the
appearance of living action.
Narrative is many, not one
NI is radically interdisciplinary, drawing on narrative concepts from many hu-
manistic fields of study. Narrative is not a single entity or a single, tightly re-
lated set of concepts. As the term is used in humanistic discourse, narrative can
mean many things. Narrative can mean a tightly woven story communicated
by a strong authorial voice to an audience. Narrative can mean the internal
imposition of coherence by which a person makes sense of her life, or the com-
munally constructed group memory by means of which a group organizes past
experience. In the broadest sense, narrative can mean an entire worldview (as
in “grand” or “master” narrative). And within each of these gross distinctions,
there lie yet more distinctions. For example, within the notion of narrative as
a tightly woven, author-given story, there lie distinctions such as literary, cin-
ematic and dramatic stories, each of which has its own set of properties and
corresponding inspirations and design implications for NI researchers. Thus
narrative is a family resemblance concept, a cover term for a rich set of ideas.
The richness of narrative presents some interesting challenges for the
emerging field of NI. One challenge is to maintain open lines of communi-
cation; with so many different inflections of the concept of narrative, workers
will have to make an effort to be clear on the notion of narrative they are using
and how it relates (or doesn't relate) to other notions of narrative.
Anoter challenge will be to remain true to the richness of narrative. AI, like
the rest of computer science, tends to prefer general and abstract formulations.
Applied to narrative, this will result in the attempt to assimilate all narrative
phenomena to a single, simplified formulation. In order to build systems, ab-
straction and simplification are necessary tools. The danger lies in forgetting
