Today’s demanding business environments are asking for more….more benefits to the customers, more intelligence from mining knowledge repositories, more choices appearing in the custom search result. And the so-called ‘expert systems’ are trying to do just that. Decoding complex user intentions into readable machine language and serving up more and more relevant information. Programs are semantically analyzing ‘entities’, parsing headers and resolving fuzzy relationships to stitch together valuable intelligence that is completely unheard of. Humans never had it so easy. They do not have to sift through piles of data from thousands of sources, what needs attention or intervention is instantly made available through QA modules. Even the number of human resolutions is becoming fewer and more complex in nature, as the easier ones are often bypassed by the constantly learning machine intelligence…

But is it all that rosy? Have we arrived at the futuristic age of ‘Artificial Agents’? May be not! The realm of human user’s experience has gone through many changes and not all of them are pleasant! For example, the ‘expert system’ may throw up many unresolved issues (at least in the beginning) that need tedious resolution with no predictable way to structure them. One can not index them in a reasonable order and many issues belong to more than one unique category. Machines still can not determine user states unless explicitly informed. And therefore takes erroneous cues to trigger some unwarranted action. The advantage boomerangs as intrusiveness!

Sengers, Boehner and Gaver suggests that Experience is an interpretation or ‘meaning-making’ by the users while using complex technical systems with AI capabilities.

Similar possibilities arise from the use of Artificial Intelligence techniques that themselves actively interpret patterns of human activity and generate responses as a function of these interpretations. Such ambient intelligences are able to actively participate in human contexts, not by attempting to completely and formally model the context, but rather by participating in the context as a non-human subject engaged in the shared construction of meaning. Such systems become an “alien presence” which, through its idiosyncratic interpretations and responses, open unusual viewpoints onto everyday human activity, providing opportunities for contemplation. Such systems share commonalities with ambient intelligence, though such work tends to be more concerned with task support than with supporting rich, affective experiences and reflection.
– ‘Experience as Interpretation’. Sengers, Boehner and Gaver

This ‘Alien presence’ is expressed through system states and most often uses ’signal detection’ and other cultural and contextual factors for co-interpreting meanings. Many times the control of the intelligence or the language of human and system interaction is exposed to the user to manage, e.g. Rule building and inspecting.
Yet users’ experiences are by necessity radically subjective, deeply dependent on
irreproducibles such as context, mood, and past personal history. And most importantly demands certain deliberation of intent that are almost beyond the rational/logical realm of the system engineering.