Everyware, Findability, and AI (Part 3)

As Part 2 in this series indicated, my interest in ubiquitous computing started with the sort of issues raised by Lucy Suchman’s initial research on artificial intelligence applications, specifically expert systems. I’ve been waiting to read Lucy’s second edition of Plans and Situated Actions, titled Human-Machine Reconfigurations before finishing this series of entries. It is an interesting read, and I think several themes introduced by Suchman’s most recent work nicely highlight the contributions in Adam Greenfield’s Everyware.

Everyware offers a number of interesting and provocative insights into the phenomena of ubiquitous computing. The most sensible, and provocative, insight offered by Greenfield relates to whether the design of ubiquitous computing needs to aim for seamless interaction with people using connected devices, or whether a rigorous focus is needed on how to make seamful interaction the guiding design practice. A seamful design approach allows people to manage the boundary conditions between ubiquitous computing and routine cultural practices. Indeed, Greenfield advises, “seamlessness must be an optional mode of presentation, not a mandatory or inescapable one” (p.238). With that proviso in mind, Greenfield offers five guidelines for experience design in ubiquitous computing applications. He presents them as “ethical” guidelines.

However, the guidelines offered by Greenfield involve more than merely ethical considerations. They speak directly to the willingness of people and organizations to use the technologies and incorporate them into everyday routines. As Greenfield himself notes, an unaddressed disconnect exists “between the current discourse around ubiquitous systems, and any discernible desire on the part of meaningfully large populations for such systems” (p. 191). The guidelines offered by Greenfield include the following design defaults:

1. Default to harmlessness: “Ubiquitous systems must default to a mode that ensures users’ physical, psychic, and financial safety” (p. 235).
2. Default to self-disclosure: “Ubiquitous systems must contain provisions for immediate and transparent querying of their ownership, use, and capabilities” (p. 237)/
3. Default to conservation of face: “Ubiquitous systems must not act in such a manner as would unduly embarrass or humiliate users, or expose them to ridicule or social opprobrium, in the course of normal operations” (p. 240).
4. Default to conservation of time: “Ubiquitous systems must not introduce undue complications into ordinary operations” (p. 244).
5. Default to deniability: “Ubiquitous systems must offer users the ability to opt out, always and at any point” (p. 246).
   

Greenfield’s guidelines point to limitations in visions of ubiquitous environments where people, devices, and computation are seamlessly integrated into everyday life. The vision of such seamless interfaces requires sensors, actuators, wireless networks, and ubiquitous devices powered by intelligent computation to blend into future living environments for people, as an invisible infrastructure. Suchman’s Human-Machine Reconfigurations characterizes the effort as a utopian dream: “That is the fantasy of the perfect, invisible infrastructure: in this case, one that joins together the promise of intelligent machines with the needs of a service economy” (2007, p. 217). Greenfield characterizes the overall quest as AI-hard, “as challenging as the creation of an artificial human level intelligence” (p. 231).

The quest for seamless interfaces for ubiquitous applications trades on the same sleight of hand as many artificial intelligence applications in the past. It is a promise that the designers and developers can substitute the agency of ubiquitous devices for the agency of people. There is no doubt that devices act as agents, nor is there any doubt that the agency of devices makes sense only when considered in the context of human activity. Each point of view typically fails to maintain perspective on the limits of both types of agency. It is easy, some might say analytically convenient, to slip into one or the other frames of reference. Suchman characterizes the point well noting that,

“I would propose that the price of recognizing the agency of artifacts need not be the denial of our own. Now that agencies of things are well established, might we not bring the human out from behind the curtain, so to speak, without disenchantment? This requires, among other things, that we acknowledge the curtain’s role. Agencies…reside neither in us nor in our artifacts but in our intra-actions” (p. 285)

Advocates of seamful design in ubiquitous computing applications implicitly recognize the point made by Suchman. She points out the obvious in an elegant way: humans design, develop, and implement ubiquitous computing. Moreover, the agency in such devices merely echoes their human creators rather than engaging in the “contingent coproduction of a shared sociomaterial world” (p. 23). In other words, ubiquitous computing devices integrated into a seamless interface must act on the engaged participation of people in everyday activities. Yet, human agency involves engaged participation with others in a manner that “requires an autobiography, a presence, and a projected future” (p. 23). As Greenfield observes, the challenge is AI-hard. Yet, a fair reading would probably characterize it as AI-impossible.

First Posted January 3, 2007.

Copyright © 2007 by Larry R. Irons

Posted by: larry on 5/8/2008 10:02:25 AM , 4 comments