![[Home]](http://meatballwiki.org/meatball.gif)
FunctionalAgent
MeatballWiki | RecentChanges | Random Page | Indices | CategoriesStandard AI definition of an agent relies on interaction, because that's how the testing of their work takes place. FunctionalAgents don't follow that rule, nor does life. Government agents are agents of the government, enforcing the will and providing information for the government. Similarly, Travel agents work as your person providing a means of providing the requirements for travel. Whether that person talks to you means something about whether that person will remain your travel agent, but it is the function, not the interaction that defines their role.
Similarly, a FunctionalAgent is something that functions for you. Under UNIX systems, ProcMail? serves as your mailsorting agent, imposing your will upon the incoming mail stream. Similar code for providing you current information is easy to code.
I tend to use the term StupidAgent to refer to this class of program, to disassociate it strongly from the class of ArtificialIntelligence programs called IntelligentAgent, SunirShah made a point recently pointing out that, assuming a point where IntelligentAgents exist with any real meaning of intelligence or sentience, forcing them to do the sort of gruntwork a FunctionalAgent/StupidAgent is required to do would be considered cruelty.