Computers That Understand What You Say

A smart-phone that engages in conversations is the next “Incoming” perturbation that will dictate how DoD will have to revise its information management practices.

DoD planners will have to include in their investment programs the availability of tactical consumer radios costing less than $300. The first firm to launch such technology is Apple with its iPhone 4S. It offers intelligent conversational capability. No other consumer computer firm has ever offered that before. We can be sure that other vendors will follow with similar products.

The iPhone 4S device is the first device that offers a reasonably capacity to perform natural language processing using semantic methods. Apple has applied computational linguistics to make it possible for the conduct of unstructured verbal and text exchanges to take place between computers and humans.
The application that does that is called SIRI. It depends for its capacity to talk back on semantic software that depends on its linguistic capability by extracting the meaning of word from the Apple cloud. Though SIRI still has problems responding to unusual requests, there are now a huge number of programmers who are enhancing the vocabulary of interactions while SIRI keeps “learning” from millions of conversations.

Over the past 20 years there have been many attempts to endow computers with a conversational capability. This involved the use of complex and very expensive special purpose hardware and software. What makes SIRI different is its reliance on packaging into a combination of conventional as well as innovative features that makes it possible to engage in simple conversations. The shirt-pocket sized iPhone include not only fully featured e-mail, office applications, calendars and an unlimited number of business applications but also a camera, a video recorder, GPS, geography-tagging, a compass, a gyro, a proximity sensor as well as face identification features.

Apple packed into a 4.9 oz. device UMTS/HSDPA/HSUPA (850, 900, 1900, 2100 MHz); GSM/EDGE (850, 900, 1800, 1900 MHz); CDMA EV-DO Rev. A (800, 1900 MHz); 802.11b/g/n Wi-Fi (802.11n 2.4GHz only) as well as Bluetooth 4.0 wireless. This makes the iPhone a software-defined radio, which covers a spectrum of frequencies. It can be encrypted for security protection.
SIRI will talk in US and UK English (U.S.), Chinese (Simplified), Chinese (Traditional), French, French (Canadian), French (Switzerland), German, Italian, Japanese (Romaji, Kana), Korean, Spanish, Arabic, Catalan, Cherokee, Croatian, Czech, Danish, Dutch, Estonian, Finnish, Flemish, Greek, Hawaiian, Hebrew, Hindi, Hungarian, Indonesian, Latvian, Lithuanian, Malay, Norwegian, Polish, Portuguese, Portuguese (Brazil), Romanian, Russian, Slovak, Swedish, Thai, Turkish, Ukrainian, Vietnamese.

DoD planners can view the iPhone 4S as a harbinger of a revolutionary new approach how people will interact in the cyber sphere. Other manufacturers will be entering into a new technology race. The issue will be which of the many competing public clouds can support their respective devices with a superior capacity to conduct intelligent conversations without delays.

DoD information architecture will have to start adopting systems that will support person-centered applications. Though business applications may remain operating in the existing mode for a time, natural language applications should be focused on meeting the warfighter’s tactical needs. New systems should be able to offer the capacity to:

• To recognize the context of commands;
• To cope with inquiries that ask for summaries of complex data;
• Respond to silent texting, without keyboard inputs;
• Allow for terse communications about missions and objectives;
• Combine GPS, geography and intelligence information;
• Deliver situational awareness to individuals;
• Collect photo and video intelligence;
• Connect to diverse applications to obtain instant answers;
• Recognize diction characteristics of a sender;
• Use face recognition as means for biometric identification;
• Deal with multiple frequencies make it a software defined radio;
• Handle multiple languages for automatic translation of conversations;
• Track all communications and assign identity to an individual.

All of the linguistic intelligence of SIRI-like devices will remain, for several decades to come, on central clouds that house petabytes and even exabytes of semantic relationships. This must be available in real-time.

Semantic methods depend on an examination of millions of sentences to extract from communications relationships between the syntax of questions and the mostly likely context in which a word or a sentence have appeared before. This requires the uses of extremely fast parallel computers that will have to subdivide the task of finding the right answers.

To maintain a 100% reliable connection between local cell-phone devices and the central repository of semantic intelligence, DoD will have to depend on the availability of a multiplicity of “on the edge” servers. This is especially necessary in the case of deployment of expeditionary forces.
The availability of personal communicators that can hold conversations is a major breakthrough in the evolution of computing. Time has come for DoD planners to prepare for that. Intelligent communications will require different data centers and different networks.

NOTE: Originally published as "Incoming" in AFCEA Signal Magazine, 2012