Chapter 1
Introduction
Content
The focus of this report is on artificial intelligence (AI) and human-computer interface (HCI) technology. Observations, conclusions, and recommendations regarding AI and HCI are presented in terms of six grand challenge areas which serve to identify key scientific and engineering issues and opportunities. Chapter 1 presents the panel's definitions of these and related terms. Chapter 2 presents the panel's general observations and recommendations regarding AI and HCI. Finally, Chapter 3 discusses computer science, AI, and HCI in terms of the six selected "grand challenge" areas and three time horizons, that is, short term (within the next 2 years), midterm (2 to 6 years), and long term (more than 6 years from now) and presents additional recommendations in these areas.
Selected Definitions
The scientific and engineering disciplines that make up or are closely allied with AI and HCI are of relatively recent origin. They date back to the 1950s at the earliest, with the first encyclopedia article on AI appearing in 1967. AI and HCI have been rapidly changing and expanding both in intellectual content and in application, especially in the last decade. The recent accelerated pace of change has been due in no small part to the almost breathtaking innovations and product developments in the supporting microelectronics, electrooptics, and display technologies—all hardware-intensive areas.
Growth in the amount of new terminology and technical jargon always accompanies such changes and advances in technology. Recognizing the futility of attempts to craft comprehensive definitions in light of these rapid changes, the panel has opted to provide brief descriptions of four frequently used terms: artificial intelligence, human-computer interface, virtual worlds, and synthetic environments.
Artificial Intelligence
Artificial intelligence is the collection of computations that at any time make it possible to assist users to perceive, reason, and act. Since it is computations that make up AI, the functions of perceiving, reasoning, and acting can be accomplished under the control of the computational device (e.g., computers or robotics) in question.1
AI at a minimum includes
- Representations of "reality," cognition, and information, along with associated methods of representation;
- Machine learning;
- Representations of vision and language;
- Robotics; and
- Virtual reality (defined below).
Human-Computer Interface
Human-computer interface (HCI) consists of the following:
- The machine integration and interpretation of data and their presentation in a form convenient to the human operator or user (i.e., displays, human intelligence emulated in computational devices, and simulation and synthetic environments).2
- The bidirectional communication of information between two powerful
- information processors: people and computers. Information can be in the form of data, symbolic knowledge, or control specifics.3
Virtual Worlds
A virtual world, or virtual reality, is a precise re-creation of a real-world environment via multisensory data and computer graphics that allows interaction between humans and synthesized objects. It consists of a set of multisensory devices employed as both actuators and effectors. "Virtual" is often used synonymously with computer-generated or synthetic.
Synthetic Environments
A synthetic environment is a reconstructed multipurpose environment with a mix of real and computer-synthesized (simulated) objects under computer control. It allows interaction between combinations of real and synthesized objects. A synthetic environment consists of a digital and analog representation of a physical environment with specified fidelity and complexity and is scalable to any size and degree of complexity.
Grand Challenge Areas
The grand challenge areas selected by the panel should be understood in the context of the usage of that term in the High Performance Computing and Communications (HPCC) 1991 Initiative of Alan Bromley, President Bush's Science Advisor, stated as follows: "The HPCC Program is driven by the recognition that unprecedented computational power and capability is needed to investigate and understand a wide range of scientific and engineering grand challenge problems. These are fundamental problems whose solution is critical to national needs."4
The panel tailored this concept to its work in the following manner: Grand challenge areas are those fundamental problem areas to which the application of scientific and engineering resources will yield much-needed improvements in capabilities and performance. They also serve to identify key scientific and engineering issues and opportunities.
In selecting the grand challenge areas, the panel applied the following two constraints:
- Only leading-edge technologies were considered, and
- Research and technology application communities had to believe that the grand challenges were susceptible to resolution and that their resolution would provide demonstrable value-added to nontrivial user groups.
The panel chose the following six grand challenge areas:
- Representation and modeling of complex systems,
- Collaborative problem solving,
- Machine learning and adaptive systems,
- Reasoning under uncertainty,
- Virtual worlds (reality), and
- Neurophysiological models of cognition.
Panel members concurred that organizing their observations and findings concerning the NRL priority topics in the terms of reference (see the preface) according to the grand challenge areas selected would bring focus to their work and would permit easy presentation of highlights, issues, and high-value applications. Table 1.1 summarizes the relationship of the NRL priority topics to the grand challenges.
Table 1.1 can be read as follows: Software production, particularly for increasingly complex systems, presents challenges in the extensive collaborative and sometimes
Table 1.1 Linkages Between Grand Challenge Areas and NRL Priority Topics.
GRAND CHALLENGE AREAS |
NRL PRIORITY TOPICS |
Representation and modeling of complex systems |
Software production |
Collaborative problem solving |
Interface technology, speech synthesis/recognition |
Machine learning and adaptive systems |
Adaptive software, speech synthesis/recognition, neural networks |
Reasoning under uncertainty |
Adaptive software, speech synthesis/recognition, neural networks |
Virtual worlds (reality) |
Interface technology, speech synthesis/recognition |
Neurophysiological models of cognition |
Neural networks |
concurrent problem-solving effort involved, and in realistic representation and modeling.5 Software for complex systems also presents challenges in reasoning under uncertainty.6 Adaptive software presents challenges in machine learning and reasoning under uncertainty. Interface technology develops collaborative (with the user) problem solving and virtual reality. Speech synthesis and recognition present challenges in collaborative problem solving, machine learning/adaptive systems, reasoning under uncertainty, and virtual worlds. Neural networks present challenges in machine learning, reasoning under uncertainty, and neurophysiological models of cognition.
The sixth topic in the terms of reference was facilities, and the related questions were ''What facilities are the highest priority to emphasize in furthering the unique strengths that a government laboratory brings to this field? Which facilities are the most appropriate at a university?" This topic is addressed in the presentation of each grand challenge area to the extent possible on the basis of the panel's collective experience.