The heaps of information piling up virtually on today’s computers present a major organizational problem for many people. A soon-to-be-released product called Smart Desktop, from a division of Seattle-based company Pi Corporation, aims to help people sort that information automatically and intelligently.
“People’s lives are inherently complex, and so everyone, no matter what their job description, has to break their lives into a series of manageable chunks,” says Jon Herlocker, vice president of engineering and CTO of Pi’s Smart Desktop division. “Those chunks become the context in which things get worked on.” Smart Desktop is designed, accordingly, to group information by project. But the software does far more than put all your Word and Excel documents in one folder. Normally, information is stored in separate applications: important URLs might be saved in Firefox’s Web history, for example, while e-mails are stored in Outlook. Smart Desktop unifies the information into a single view. When the system understands that a user is working on a particular project, it will bring together related files and e-mails, and it will also bring related, newly arriving e-mails to the user’s attention.
The system will make this information available to users in two ways, says John Forbes, president of Pi’s Smart Desktop division. All materials related to a project will be visible inside Smart Desktop’s user interface. The system will also tag the files so that the user can search for project tags created by Smart Desktop in Outlook, for example. “One way to think about Smart Desktop technology is automatic tagging,” says Forbes.
Herlocker says that the product is designed to work with Microsoft Windows, in order to allow a user to continue working with her existing operating system at the same time that she runs Smart Desktop. In addition to supporting popular Microsoft applications, Smart Desktop will work with applications such as Adobe Acrobat, Google Docs, and Zoho Sheets.
Smart Desktop’s underlying technology comes from Oregon State University’s Task Tracer research project, which was a part of the CALO artificial-intelligence project. (See “Software That Learns from Users.”) The Task Tracer research employed machine-learning techniques to build desktop applications that were able to learn and observe the task that a user was working on at any given moment. But Herlocker says that Smart Desktop is being careful about how it presents its artificial-intelligence roots. “If you anthropomorphize too much, people’s expectations become too high because they expect it to be as capable as a human,” he says.
While the exact workings of Smart Desktop have not yet been finalized, Herlocker says that the aim is to make the intelligence as invisible as possible. For example, the system might reorder search results to bring more-relevant results closer to the top. Or, when a user goes to save a file, the default folders will have been selected based on the project the user is working on, reducing the number of clicks needed to find the right place to store the information.
“One key is having a solution that works without making users change the way they work,” says Forbes.
Herlocker says that the company has tried to take into consideration how much people are willing to train a system, and how many interruptions they are willing to tolerate. “We’ve tried to pick the middle ground that leverages as much as possible what we know to date about machine learning and artificial intelligence, but at the same time remain very realistic about the impact of using machine-learning systems in a human context,” he says.
Pi isn’t the only company working on products of this type. Microsoft, through its Center for Information Work, is also developing a desktop assistant with similar capabilities. Currently in its prototype form, Microsoft’s desktop assistant is designed to bring together an individual’s communications, calendar, and other related information so that it creates a timeline view of her day. The system brings up relevant content based on what the user is currently doing, or what she should be doing at that time. “The software behind the desktop assistant is ‘smart’ in that it understands what information is important, and prioritizes communications and information, allowing the user to filter down the noise to a manageable level,” says Russ Burtner, user-experience designer for the Center for Information Work.
Smart Desktop is currently available in an invitation-only preview, but Herlocker says that he expects the full product to be released in late 2008.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
Data analytics reveal real business value
Sophisticated analytics tools mine insights from data, optimizing operational processes across the enterprise.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.