1. Sponsor: DARPA Augmented Cognition Program
2. Period of Proposed Research: 1 October 2001 through 30 September 2004
3. Total Estimated Cost for Proposed Period
1 April 2001 – 31 December 2001: $271,120,
1 October 2001-30 September 2004: $429,000 per year for three years
4. Investigators: Michael Zyda, Rudy Darken & Michael Capps
5. Brief Description: The DARPA Augmented Cognition Program is looking at ways in which an electronic prosthesis can be developed that augments cognitive processing. One way to think about such a machine might be that it is like the annoying help feature of Microsoft Word that tells you “I think you are trying to write a business letter. Would you like me to suggest to you how?” What we would perhaps like is something grander than help with the written page. We would like something that watches over our shoulder and provides us appropriate information (guidance, warning, help) when it senses our current “context”. We call such a machine a context machine. The work we propose is to develop a plan for constructing such a machine, a timeline for its construction and an elucidation of the technologies (sensors, computing, displays, …) required for the construction of a portable context machine.
6. Keywords: Computing, agent-based simulation, cognition, wireless.
7. DoD Key Technology Areas: computers, software
The DARPA Augmented Cognition Program is looking at ways in which an electronic prosthesis can be developed that augments cognitive processing. One way to think about such a machine might be that it is like the annoying help feature of Microsoft Word that tells you “I think you are trying to write a business letter. Would you like me to suggest to you how?” What we would perhaps like is something grander than help with the written page. We would like something that watches over our shoulder and provides us appropriate information (guidance, warning, help) when it senses our current “context”. We call such a machine a context machine. The work we propose is to develop a plan for constructing such a machine, a timeline for its construction and an elucidation of the technologies (sensors, computing, displays, …) required for the construction of a portable context machine.
The Neal Stephenson book entitled The Diamond Age - A Young Lady’s Illustrated Primer (1995) contains a very sophisticated piece of technology, the Primer. We quote a review of the text to provide an overview of the story and the role of the primer.
“Stephenson's dazzling cyberspace adventure, Snow Crash (1992), drew accolades as one of the most innovative, thought-provoking first sf novels since William Gibson's Neuromancer. Unlike Gibson, who followed with lesser sequels, Stephenson breaks new ground in a grand-scale forecast of the coming nanotechnological revolution. John Percival Hackworth is a cultured nanotech engineer who risks the censure of his neo-Victorian social class, or tribe, when he forges a copy of an interactive, computer-driven book called A Young Lady's Illustrated Primer. With the unprecedented power to single-handedly educate its reader, the primer is designed to shape the values and maintain the superiority of the dominant tribe. During a mugging, however, Hackworth loses the copy to a lower-class thug, who in turn gives it to his sister Nell. As Nell learns secrets from the magic book, her understanding of herself and her world grows in ways the primer's designers never intended, and the entire destiny of society changes irrevocably. Stephenson's command of character and stylistic nuance has grown captivatingly stronger, and he now offers startling new ideas in virtually every paragraph. With breathtaking vision and insight, Stephenson establishes himself as not only a major voice in contemporary sf but also a prophet of technology's future.” Carl Hays, Booklist, Review from Amazon.com
In the story, the character Nell reads the primer and it provides her appropriate games, games related to what is going on in her life. The primer “senses” her context and adapts its presentation and content to provide her an analogous story in the primer, one that educates her in the ways of the world, if she understands the message from the game and can make the cognitive transition from the game to the real world. The purpose of the primer is to make a proper lady out of the reader but if the primer can sense her “context” then that primer, if it existed, could be more direct and provide her warnings, guidance and help at the appropriate time. For purposes of this proposal, we rename the primer the “context machine” seeing as that its primary function is to watch over our shoulder and provide us appropriate information (guidance, warning, help) when it senses our current “context”.
There are lots of uses for a context machine provided it is lightweight enough and unobtrusive in nature such that we wear it all the time. Let’s make some assumptions for such a machine. It has a very powerful computing engine, with near infinite on-board memory and effectively infinite permanent storage. It is wireless and net connected, with high-speed access to the larger net. Its batteries are small and long-lasting. It has on-board sensors that maybe sit on your shoulder or in the earpiece of your glasses or are attached to your wrist, using Bluetooth or its descendent for on-body communication. Those sensors can record video in visible and infrared spectra, audio, smell, temperature, wind, radiation, heart rate, skin temperature, and skin surface moisture. The machine knows where you are in the world using on-board GPS. It knows your body’s spatial orientations and configuration. There is software on-board that can process the sensor information in real-time, such that it can immediately make use of the sensed information in abstract fashion and record context in real-time.
So we kind of have an idea of what this machine has in it technologically. What would we use it for? Well, there are lots of things that it could be used for but imagine that we would like to have such a machine to support the transition of jobs for high turnover, very important positions. We would additionally like such a machine to provide education and background to people, in context, for job and watch transitions. Maybe we might use the context machine to capture the knowledge of warfighters from Desert Storm before they are all out of the service over the next 10 years. So there are lots of real reasons why we would like such a machine.
Excellent question. With this white paper, we propose to look into the hardware and software available now and in the near-term, say over the next 5 to 10 years, so that we can outline the parameters of what we can do with respect to building such a machine. We will identify available component subsystems that can be leveraged, subsystems that can perhaps be modified for our purposes. Our deliverables will be:
1. A report on the component hardware and software pieces for the context machine. That report will describe such components with respect to what is available now, making the assumption that such pieces will be integratable into the final machine in some way.
2. A timeline of availability and potential availability for the component hardware and software pieces.
3. A research agenda listing the component pieces that may not be available but which are essential for the construction of the context machine.