Skip to Content

Searching for Intelligent E-mail Agents

Beyond spam filtering, tools emerge to cope with e-mail overload.
August 8, 2011

E-mail may have been the Internet’s first “killer app,” but keeping up with it has become sheer murder. The volume of e-mail defies comprehension: by one count, 32 billion messages a day were sent in 2010, a figure that does not include the roughly 90 percent of e-mails that are spam.

Get organized: Google can automatically prioritize important e-mails.

A growing number of products and research efforts aim to ensure that e-mail overload doesn’t cancel out the productivity-enhancing benefits of IT. 

Google’s Gmail “Priority Inbox” is an early effort in this direction. The feature adds a special icon to messages it judges to be important, allowing users to tend to them first.

It judges by a number of factors. For one, it looks at the sender of the message and the time it was transmitted, which can offer clues to the e-mail’s importance: a message your mom sent you at 2 a.m. is likely to be more important than one she sends at noon.

Gmail also “reads” the message itself, and when it spots a term like “ASAP” or “urgent,” it is more likely to move that message to the front of the line. It uses numerous other clues—for example, an individual message to you from your boss is presumed to be more important than group messages.

Just as it is with the details of its main Web-search algorithm, Google is mum about the specifics of the code that determines whether or not a message is important. But Alex Gawley, senior product manager for Gmail, says the program, by watching the user’s behavior, increases in accuracy over time.

For all its advantages, Google’s Priority Inbox demonstrates how hard the e-mail overload problem really is. Many Gmail users say that while the software does an acceptable job of separating e-mail wheat from chaff, even their Priority Inbox quickly fills up with items demanding attention.

One of the reasons for that, say software researchers, is that e-mail programs remain “dumb.” There are a number of common actions that users may need to take after reading an e-mail, such as updating a piece of software or entering an item into a calendar. But current e-mail programs offer limited help in performing those tasks.

Some companies are developing “smarter” mail programs to help out. Farzin Arsanjani, president of HyperOffice, in Rockville, Maryland, says his company’s collaborative software is designed to break down barriers between current office-productivity programs, enabling information to flow more easily between, for example, Outlook and Excel.

What about an e-mail program that is smart enough to actually answer mail for you? Such things exist, but only in very limited domains that are irrelevant to the average problems of an everyday office.

One example is Project Radar, a four-year, multi-million-dollar DARPA effort to develop automated e-mail reading that would be useful to the armed forces. Scientists at Carnegie Mellon and SRI International collaborated on the work and eventually produced a system that could handle a small group of logistical and scheduling messages.

SRI eventually spun off from Project Radar “personal assistant” software known as Siri, which it later sold to Apple. Siri supplies mobile-phone users with answers to relatively simple questions from well-structured databases, such as “Where is the nearest post office?” and “What restaurants around here are still open?”

Michael Freed, program director at SRI’s Artificial Intelligence Center, envisions future personal assistants that are even more robust than Siri, including ones that can begin to respond to the endless possible scenarios contained in the average e-mail inbox. “But they are not going to be perfect. If they catch things only 90 percent of the time, is that going to be good enough for you? A lot is going to depend on how much people will tolerate them when they don’t work right.”

Considering how much time most people now spend on e-mail, the odds are that tolerance will be relatively high.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

The problem with plug-in hybrids? Their drivers.

Plug-in hybrids are often sold as a transition to EVs, but new data from Europe shows we’re still underestimating the emissions they produce.

How scientists traced a mysterious covid case back to six toilets

When wastewater surveillance turns into a hunt for a single infected individual, the ethics get tricky.

Google DeepMind’s new generative model makes Super Mario–like games from scratch

Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.