Wednesday, April 20, 2011

Paper Reading #25 - Email Overload

Comments:
Comment 1
Comment 2

References:
Title: Agent-Assisted Task Management that Reduces Email Overload
Author: Andrew Faulring, Brad Myers, Ken Mohnkern, Bradley Schmerl, Aaron Steinfeld, John Zimmerman, Asim Smailagic, Jeffery Hansen, and Daniel Siewiorek
Venue: IUI 2010, Feb. 7-10 2010

Summary:
In this paper, the authors describe a new mail system that uses AI to divide e-mails into a selection of tasks. They then show that this very different method provides positive results.

They begin by describing the intricacies of the task system. When e-mails enter the inbox, an AI assistant parses the e-mails and tries to figure out what classification the task should lie under. Then, it chooses to either place it into a classification or place it in an area where the user can choose.

In addition, the e-mail client also provides a scheduling interface, which also includes an AI assistant. The AI assistant looks through the e-mail tasks and assigns what it believes to be a good amount of time for each, and prioritizes the user's future schedule. The user can then choose what tasks they are working on.

They then show the results of using this system on productivity. People using this system with the AI task assistant and e-mail assistant tend to get more meaningful tasks done than those who do not. With only the e-mail assistant, users get more overall tasks done, but they get less important tasks done.

Discussion:
I was actually quite excited about this research. The idea of having a small AI assisting me with my tasks seems like a really cool, sci-fi idea. Additionally, even at this stage, it seems to be working well, so I hope they can actually bring this to market soon.

One concern I have with the software is that they do not describe how configuration will work. I am curious if in a final design if there will be editable categories, or if because of how the AI works, there will only be preset task categories.

(Image courtesy of: Download Software)

4 comments:

  1. I can see how something like this would be useful for organizational purposes. I didn't actually read this paper so I'm not sure how accurate this is, but I would be concerned that it would classify emails wrongly, leading to confusion or lost emails.

    ReplyDelete
  2. I agree with Shena, I was wondering about how if this was used in an environment where massive amount of e-mails are expected it would be terrible if certain e-mails were classified wrongly. Even if it was only a couple this could have a big impact.

    ReplyDelete
  3. While this is a cool idea, I think this technology already exists in the form of spam filters. Overall though, using algorithms to manage your mail classifications can be dangerous. Every once in a while they get it wrong, and classify something important as junk. This can have pretty serious consequences depending on the content of the email you missed.

    ReplyDelete
  4. I don't have enough of a problem with my email to really want to use this product. I can see how some people might though.

    ReplyDelete