Paper Reading # 17: Agent-Assisted Task Management that Reduces Email Overload

Agent-Assisted Task Management that Reduces Email Overload
Andrew Faulring, Brad Myers, Ken Mohnkern, Bradley Schmerl, Aaron Steinfeld, John Zimmerman, Asim Smailagic, Jeffery Hansen, and Daniel Siewiorek
School of Computer Science, Carnegie Mellon University

In this article the authors introduce their new system called RADAR. Essentially what it is, is a email scanning system that helps a user to organize emails in the event of email overload. When a user has fifty plus emails sometimes it can be very hard to get through them all in a reasonable time and keep them organized and their system aims to assist with that. Essentially what the system does is use scanning algorithms and searches to identify characteristics that are similar in emails and then provide them with a tag that helps organize them. This is a simple overlay for an email that adds tags and tasks to the users email and then compares them and puts them into groups. The problem with this is that the user is essentially required to read the email twice to be able to check to make sure that the email is in the correct folder and has the correct tag. This is assisted by the RADAR task-creating function that helps the user organize the task and choose the correct tag for the email. The system also has a task scheduleing algorithm that allows the user to create a plan to go through the emails and put them into the correct catagories.Users who tested this system saw its original work as being very successful and it did indeed help them get through the organization and reading of emails faster. The authors claim that the biggest task they have now is making sure that the email program is robust enough to handle real world tasks. The users did find the AI really helpful but they didn't really understand the task scheduler and were unable to make great use of it. The users found that sometimes the task suggestion algorithm made them take MORE time reading emails and some people just found it faster to do without it. The study was called a success and the authors closed by saying they just want to make the AI more robust.

This article for me was rather dry. It seems like a really good system but it seemed like something that was really excessive and in some cases I wouldn't actually find much use for it. In fact, a lot of times I am checking emails on my phone and not at my computer so even if I had this program I really question how much I would use it. It does seem like a really nice system if you are working and checking your email at work all the time and you constantly have fifty emails in your box unchecked but I think that is a very small subset of users. I think all the excess features that they added beyond the AI that "proofreads" the email and assigns a tag also seem very excessive. The idea of having an algorithm that suggests when to look at what email and gives you a schedule to read emails seems like something that I might get rather frustrated with for getting in the way and constantly providing input. Despite this I do think that this work was a big success and the authors did do a really good job with their functionality. I am very curious as to how they plan to improve the algorithm and wanted to know more about that. Also, if this is going to be marketed as a product I don't think I will be picking it up anytime soon.