Search
Program Calendar
Browse By Day
Search Tips
Personal Schedule
Sign In
The mid-1990s experienced a bubble of agents. Software agents were everywhere, or about to be, as companies, university labs, and the press hyped their coming as a solution to human time management and information overload. AI was to be the answer, as agents could be given goals to achieve and would then work autonomously on behalf of their users, while users could be off doing more important human tasks (one company suggested taking “a hike in the hills” while
your agent searched the internet for what you needed). The excitement over agents came to a head in the form a debate between two prominent computer scientists: Pattie Maes, a pioneer in creating AI agents at MIT, and Jaron Lanier, best known then for his early virtual reality work. For Maes, agents would be a new form of “support staff” that everyone could have access to and reap the benefits from. For Lanier, agents made “people redefine themselves into lesser beings.”
Both foresaw change, positive and negative, on the horizon, although the bubble burst on agents before they could take off as a consumer technology. This talk examines what could have been: the possibilities for AI agents in the 1990s, contrasting views on how they might alter society, and some of the attempts to actually create and use them. Their role as servants was widely discussed, and yet these were servants that could operate without your knowledge or even
awareness. They could be programmed to achieve goals that users had never set, such as monitoring travel conditions for an upcoming trip without being asked to do so. I find a further conflict, not just in the effects agent use might have on human lives but in user versus programmer authority over what was best for the user. Agents were sometimes caught in between the desires of different humans: their creators and their users.