Autonomous Interface Agents

Web applications that are context aware will be able to make greater use of autonomous agents to directly manipulate graphical objects and affect the users display.  The MIT paper titled "Autonomous Interface Agents" says of autonomous agents:

An autonomous agent is an agent program that operates in parallel with the user. Autonomy says that the agent is, conceptually at least, always running. The agent may discover a condition that might interest the user and independently decide to notify him or her. The agent may remain active based on previous input long after the user has issued other commands or has even turned the computer off.

How many times have you built an application and then, when speaking to a user about how they use it been confounded to learn that they don't know about some of the best little features that you included in it?

So through better use of agents, we can envisage a system where users are offered cues to increase their productivity with a tool.  Some simple illustrations of this might include systems which:

  • know which features a user hasn't used and offers some ad-hoc advice about that feature to a user.
  • can offer help whilst a user is using a particular feature
  • remember past screens that a user has used and offer them as "favourite" short-cuts
  • offer quick-fix solutions for filling-in forms with valid data

Autonomous Interface Agents


  • Reading this reminds me of BonziBuddy and Clippy, which exemplifies this on the "annoying" end of the scale. I'm sure there is a noble use for this concept, but in today's world of total user control, spyware, and etc., it is more likely to be shunned by end-users. A niche implementation would probably be more successful, (eg. agent that helps disabled users browse inaccessible web pages).

  • Hello anonymous Vurg...

    When thinking of modern agents think more of how things such as the autocorrect options work in the new versions of Office and try to leave Clippy in your past. Elsewhere in that series of papers the authors describe the following scenario as an example of the expense of explicit actions being performed by the user:


    The reason to move away from the black-box model is that we would like to challenge several of the assumptions that underlie this model. First is the assumption of explicit input. In user interfaces, explicit input from the user is expensive; it slows down the interaction, interrupts the user's train of thought, and raises the possibility of mistakes.... in many user interface situations, the goal is to minimize input explicitly provided by the user.


Comments have been disabled for this content.