A talk at an international gathering of the World Future Society in Washington, DC.
The onset of the Information Age, the latter half of the 20th Century, has been stimulated by the science of communications and computers, of signals and sensing systems, whose impact on societal crisis or conflict is yet to be discerned. The history of civilization so far has naturally been preoccupied with use of ignorance, with the exploitation of situations in war, governmental control, political moves which come from the concentration of information, or the manipulation and distortion of it as in propaganda of dictatorial and collectivist societies. Thus, the new issue is how crises and conflicts may be made and acted on in a modern milieu, in which information is widely or universally available, rapidly verifiable, but nevertheless subject to even more dramatic controls and manipulations than in earlier eras (the various communiqués of the recent Chernobyl nuclear events provide dramatic illustration). This remains to be assessed by such programs as “Future Focus” of the World Future Society. We shall seek to sample in this report the dimensions of change in these matters as influenced by communications and information handling as well as some related technologies such as robotics and automata.
In realizing the drastic shifts of speed, capacity, and scope in this realm, which are assured for the next 15 years, we shall see that they affect deeply our basic perception of civilized living on the planet. It is likely that these international changes will deeply influence our perception of crises and conflicts, and how they are described and defined. Thus, we can not just apply modern schemes of electronics and photonics and computers and networks to the present ideas of what such critical phenomena are. Rather, we should probe the effects of changes in knowledge of waves and particles, of systems and technical situations, on humanism and the qualities which shape communities and nations.
Examples of these basic shifts in frame of reference of the people in nation states appear in such matters as the present and growing world food surplus. There the attentions, energies, and social values presumably dominated since the Stone Age by hunger and seeking assurance of palatable food supplies are extensively being displaced by other socio-economic desires, even though still intensively consumer oriented. Style, fashion, amusement venturesomeness such as in sports and gambling, etc., are playing larger and larger parts in human wants. Indeed, they evolve as a culture shifts, into what will probably soon be called human needs. Crises depending on modifications of these wants and needs will thus bear different relationships to information than in conventional times. Then, folk knowledge and its distribution governed not only the arts of agriculture and feeding, but also many of the habits of government and institutions. Their fluctuations, in turn, augmented often by natural disasters, weathers, earthquakes and the like, culminated in crisis generation.
So we trace now a circumstance which will doubtless recur often in one form or another in this 1986 conference and in the years to the end of the century which it signifies. Namely, perturbations in the evolving system of information and communications may themselves generate needs for crisis management. Just as they offer important resources for crisis management, these concurrent causal factors need to be recognized and indeed capitalized in the years ahead.
For instance, what constitutes equity under Law and Civil Rights in the conditions where the liabilities of influencing or of damaging another's perception arise? What is the ethic of shaping of information or shrinking it, without either falsehood or actual withholding? These old queries take on a new slant as the access to knowledge widens and speeds up. What kind of modifications in the age-old democratic system of delegating authority on behalf of the whole community, the firm, or the institution will come in when the constituents doing the delegating are themselves increasingly capable of dealing directly with knowledge-based decisions? These ancient issues used to be based on an exercise of common sense or more generally detachment from the decisions. But we are now growing a society unsatisfied with that. Accordingly, the endless frontier of information and communication technology and science must be intimately connected with the structure of delegation and social responsibility. Particularly, this will result in a vast widening of action bases, since horizontal rather than only vertical distribution of knowledge, will give a valid pathway for accurate and tenable decisions.
So though we may be moving into a future when a shortage of information, a prohibition of familiar knowledge, a sudden limitation on the simple graphics, less literate inputs to the citizens' video, may themselves generate new kinds of crises, as well as dramatic opportunities. We have celebrated the Communication and Information Age since the mid-Century as a resource for the support of liberty. Indeed it represents an epic extension of the people's literacy, which we have expanded so intensely for a century, and then allowed to sag in this last decade. Nevertheless, that we must restore and then look forward to these new opportunities, arising from the global access to knowledge and organizing of knowledge for action which are transforming human life.
Let us now have in an Age of Liberty, which we have been celebrating as a symbolic Centennial in these last days, augmented by an Age of Integrity, in which information is accumulated and provided by new dimensions of communication and digital processing. Especially, in these ways our people can feel confident that their government, their economy, their physicians, their associates at work, their teachers, their lawyers, and all the rest of our complex interdependent compatriots, will actually know what they are doing. What is being proposed is a reform in the organization and distribution of knowledge for everyday life, as well as for the mounting host of professional or specialty activities. The goal is comparable to the reform in education for literacy and basic skills which is now thankfully underway in this nation.
We have adopted in current years, the term “Yuppies” for “young urban professional” with the assumption that we are making role models for the future generations of new professional talents in business, the arts, law, medicine, government service, and the like. But actually, ironically, the capabilities for handling large bodies of professional information by such an expanded group are inadequate and in many cases illusory. Indeed, we have not yet quite turned the tide for the restoration of literacy, but shall do so in the next few years, we believe. We ought then to have before us a goal of dilating the noösphere (according to Teilhard DeChardin), so that the next century will involve levels of realism and recognition in what we know and do not know which will brighten and sharpen lives of professionals, and all whom they serve.
Along with the concept that crises and emergencies are changing character, (recognizing dimensions of S&T of national size and character) especially because of the vast volumes of information involved and the crucial role of knowledge in oncoming years for almost any variations in civilized life, it is not surprising that Future Focus is worldwide. Likewise, each of these variations we speak of which are being dealt with in this conference for the next 15 years, and include matters of business, education, emergency management, environment, global debt and finance help, human values, media, new technologies, peace and security, and work and careers, -- all depend on some aspect of knowledge transmission and processing in a global network.
Likewise, the struggle for resources themselves, minerals and oil, clean air and water, fertile soil and livable cities, all depend on knowledge in detail and depth, knowledge of authenticity and accuracy far outside the experience of prior human history. So we believe that the science and technology of these times, enabling world sharing of knowledge and learning by a new communications, and world organizing and processing of it by digital machines, computers and automata, will be the way to lessen and to meet the crises that will come. Thus, in various prophecies from this Society and elsewhere of devastating nuclear assaults such as bombs from Third Countries, of earthquakes such as a slide from the San Andreas Fault or eruptions from mega Mt. Helena volcanoes, or the melting of the ice caps from carbon dioxide greenhouse effects, the total impact has to be regarded globally. But in every case, the results will crucially be shaped by the information that is obtained and applied.
Accordingly, let us examine what could be done and what is being done to give our nation and the world the best ways to establish an age of integrity of information and assure access to it. Now obviously, the first stage in this is understanding. This comes usually through some mechanism of literacy. So when the 5 billions inhabiting the planet can be expected to grasp ideas, as we have accented in the work of the National Commission on Excellence and Education reported in “A Nation At Risk …,” the most immediate and intense effort is required to enhance literacy and reasoning among our youth. We are assuming that that will happen here long before the next 15 years are up, but it needs to spread elsewhere. Although about a third of our current U.S. high school graduates are incompetent in simple grasping of ideas from language and of doing logic in arithmetic. The proportion may be higher in other nations, just as it is very much lower in Japan, perhaps much of Europe, and elsewhere. But let us look at what a relatively literate humanity could do about crises and conflict resolution, through the support of communication and knowledge handling aids. Let us look at this in a sequence of regarding a whole planetary system, followed by finer and finer detail, so as to get a feeling for what atoms and molecules eventually can be disposed to support specifically this handling of information and extension of the reach of human senses and cognition. Provision of a global network has been articulated by the retiring Chairman of the American Telephone and Telegraph Company, Mr. Charles L. Brown, in his principal address at the U.S. Telephone Association, October 14, 1985. Under the title “Universal Information Services,” Mr. Brown has reflected the technical and operational possibilities, including introduction of the Integrated Services Digital Network (ISDN), which could in one move transcend the language, distance, cultures, borders, boundaries, fragmentations of this globe. A suitable technical discussion of the ISDN itself, the kind of international agreements now being pursued by our industry and by certain elements of the Department of State, would take all the rest of our talk. It is sufficient to say that digitalization, with the vast capacities and bandwidths that we shall note presently, can lead to an era of language translation, of data expression, of graphics, capable of introducing a cohesion into human affairs utterly beyond concepts of only a few years ago. The striking thing is, we know how to do most of this now. We know that it would be an essential element in dealing with nuclear infernos, refractory viral plagues like AIDS, earthquakes and typhoons. So we confront an exercise in futurism in which already early forms, such as our hotline between Washington and Moscow, have demonstrated values.
So we know that the microwaves from satellites and the photonics fibers are ready to join all communities across all seas and throughout all continents. In the Atlantic, our TAT-8 and soon to begin TAT-9, will by 1990 crosslink the Atlantic community. TAT-8 has a capacity of 592 megabits per second and competing companies such as Submarine Lightwave Cable, and Tel Optik, emphasize particularly video transmission and claim a new offering of 6.7 gigabits per second, coming from 12 fibers per cable, each supporting 560 megabits per second. Thus is enabled a capability of 144 broadcast quality video channels of 45 megabits per second apiece, or 72 such video channels and more than 1,000 data channels at 1.544 megabits per second. Similar facilities are in progress for the Pacific, and the extraordinarily flexible branching qualities of photonics cables assure spread throughout the Mediterranean the Indian, and many other ocean areas. These will connect easily with onshore facilities, both analog and digital. British Telecom believes that half of its long distance telephone routes will be photonic by 1990.
The science of optics behind these systems, beginning with the discovery of the laser by Schawlow and Towne in 1958, is one of the superb epochs of basic discovery being translated to useful systems. Graded index fibers, single mode fibers, and lightsource changes permit unprecedented spacings of 30 to 60 miles between repeaters, already. That distance is growing, with the prospects of oxide and halide based fibers being even better than presently superb silica.
Indeed, the discussions ongoing in CCITT for an ISDN standard themselves advance global interests and commonality. At present, however, the USA capacity bandwidth interface is on the existing T1, with 1.544 megabits per second, whereas in Europe the primary high-frequency interface will be the present 2.048 megabits per second. But already all regions may agree on a 144 kilobit per second basic rate, which will keep the more modest sized users in world tune and harmony. Not coincidentally, the European communities’ ESPRIT information technology research program, recently attested to in President Mitterand’s visit to the U.S., is active in this set of standards. Likewise, the communication satellites, beginning with the Echo and Telstar experiments of the 1960's, succeeded by the Intelsat of 1965, have continued to expand global coverage. Although the last 6 or 7 launches have failed, the present satellites are being modified in design and the entire 16-satellite system couples with new continental satellites such as AT&T's Skynet Digital Service so that growing flexibility is assured. Further, local area networks (LAN's) are dramatically augmenting interconnections in digital systems of computers and communications. These represent again in such forms as Starlan, new ways of coordinating business, government, educational and social activities through common and creative facilities.
Let us now move to another extreme of the future interactions of people and nations that may be central to crisis and conflict resolution. For now it is established and realized that there is going to be a network which can interrelate and combine the interests and habits of humanity. We should next think of what general characteristics of knowledge, especially of integrity in information, should be exercising the network. One prime element is the approximations which knowledge of people and nature demands. New universal methods such as Karmarkar's, Algorithm, which can be computed with large capacity machines, use one appealing approach to approximating the best solutions of complex problems. These range widely, from optimal arrangements of telecommunications in the Pacific basin, to the optimal distribution of gasoline stations serving Philadelphia or Panama City to processes for immunization of an entire population under siege of plague. The same kind of approximations are needed for road, food service and other resources. Now beyond that, however, as we have pointed out in a radio colloquy with C. P. Snow preceding his return in the final years of his life to the United Kingdom, absolute and final truths are hard to come by in science or anything else. We should use the next years as ways of approximating, of averaging, of assessing situations which may modulate the culture of finality and absolutism in very important forms. This can notably improve giving suitable information on crises or on “normal” affairs.
For instance, our Director of Mathematical Research, Dr. Ronald Graham, and his co-worker, statistician Dr. Persi Diaconsis of Cal. Tech., emphasized that there are many applications of Radon transform that indicate whether averages and approximations of that kind are suitable. Special averages may show up as patterns in complicated data bases that are used to assess some situation, like the spread of disease or the shifts in climate or special precipitation, or whatever. As these “projection pursuit” methods are being applied, likewise they are revealing what conditions are required from these projections so that a continuous function may be reconstructed. An example is in x-ray tomographyhow the x-ray beam, short of infinite number of them, would have to be used to reconstruct the tissue cross sections of human patients.
Now this approximating idea is symptomatic of new ways of thought on which global decisions are often basedand which in turn should be supported by modern machine and communication networks. Of course, should be underlying them, must be a special assurance of quality control, reliability, not only of the information sources and networks but of the “facts” themselves. Again, mathematics and statistics can spread this around, so that the critical variables which assure stability, authenticity of behavior of systems, especially sensing and communications systems, can be greatly improved. Again an age of integrity will gain.
But despite these valuable supplements to human behavior, we must expect the human element to dominate through speech and text. We have already indicated the central role of language in information interchange. It is especially important, in addition to machine translation aids which have moved slowly but steadily in recent years, to have a further connection of speech quality with the structure of sentences and ideas to be communicated in the global web.
These organic interfaces are included in the general realms of human signaling (primarily speech, hearing, and vision) and of human cognition including memory, thought, reasoning, and the like. The latter functions are presumably implied in fashionable terms such as artificial intelligence or AI, which is produced by various combinations of software and hardware in computers and automata. The foremost modes of direct communications by speech, hearing, and vision as well as gestures and facial expressions (perceived visually), are of course the essence of telecommunications and other transmissions. The era of information integrity that we are proposing should, however, apply new dimensions to these technical practices. Again, as in the other areas already noted, basic science and invention are providing such dimensions. Considering the organic interface itself, where machines couple with voice, hearing, and vision, and maybe even tactile feeling, are paramount, many years of-research on speech and hearing are leading to speech recognition. This involves direct inputs to machines and automata, and then indirectly to ingenious restructuring of prose and improvement in intelligibility. The latter is compelling, for nowadays we need, in dealing with crises and large scale actions, more intelligibility to match the very large intelligence flow itself.
Accordingly, we should apply promptly in preparation for the future, things we have learned about language structure and clarity of expression through achieving machine voice synthesis. We should apply these in early stages of teaching and educational reform for literacy. We should use these skills especially in governmental communications with large communities, as in dealing with emergencies, military affairs and the broad realm of information supplies of the kind discussed earlier.
Thus, the pioneering work in our Laboratories of Coker, Mrs. Umeda and others is finding already a strong engineering use. But it ought to be extended into popular communication, in preparation for near term, emergency utilization, but also for the future operation of all manner of machines for word processing, document preparation, computer operation. Ultimately will come new facility of voice and thus human-guided manufacturing service in the general economy. While these voice interfaces are unduly speaker dependent now, major gains in accounting for this are in progress. Some enterprises such as that of Speech Systems Incorporated is concentrating on phoneme recognition, believing that the 40 known phonemes can be used for synthesis more efficiently than the millions of individual words in English or other languages. Another aspect of this approach applied by Speech Systems Incorporated, and also by Kurzweil does special sound spectrum analyzing, that either accounts for or discounts individualities. Where relatively small vocabularies are involved, template matching is also tried using special parameters of particular words derived from energy-frequency-time parameters. These are in turn related to the linear predictive coefficients effectively used by our associates in modeling the human local tract, for outputs we have demonstrated some years ago.
Already, the automobile industry spoken commands on the shop floor are guiding some important machine activities. But the main point is that we would see basic redirection of socio-economic and governmental processes when the human voice input, with all of its nuances and adaptability as long recognized in the telephone system, becomes immediately effective for digital machine control. Further, all kinds of interesting new techniques are in process in the research on speaker-independent machine interfaces for matching and identifying individualities in speech that are far beyond the speech spectrograms characterizations of some years ago. For instance, in our labs Steven E. Levinson is applying models of evolving Markov chains, through dynamic time warping, when a given signal is purposely altered to align its parameters with those of a reference group. When this warping is then minimized, additional flexibility in recognition is found. Michael Brown of Bell Laboratories has already operated a 5 Cartesian axes robotic arm by voice. The arm then determines its position and interactions by ultrasonic range finding. With 200 word vocabulary speech, the robot can be well trained to pick up and move back and forth a particular object which has been described in words to it.
Also we have said, these drastic shifts in person/machine interface organically will lead to equally profound advances in intelligibility, and especially in document output, with machine aids. We have for some years been using locally a derivative of the UNIX program for text analysis and synthesis. It is now generally available under the name “Writers’ Workbench.” It was created by a combination of linguists, computer scientist and psychologists, in order to clarify the large volumes of prose we generated internally and to give our scientist and engineers a rapid feedback of how they were slinging the adverbs and adjectives around. Some 23 programs were generated, that did style checking with respect to sentence links and structures and parts of speech which could be recognized automatically. Dr. Lorinda Cherry was a key computer scientist and Dr. Nina McDonald, the special psycholinguist, who achieved this gain in making intelligibility a fashionable affiliate of machine processing. First at Colorado State University in English Department and now at many other places, we are seeing our ambitions expressed in the lecture of some years ago being realized in prose whose precision and intelligibility indeed felicity, are astonishing. Although the best exercise of how to write understandably presently requires 35 to 40 specific programs, adding to about 700 kilobytes of memory, major aids to spelling, punctuation, capitalization, copy editing, are also provided in reflecting style and reporting it to the writer. About 70 variables are used in estimating readability, based on the number of sentences, word and sentence lengths, simple or compound type, passive voice, occurrence, and the like. At IBM, the Epistle Project is applying sentence parsing for similar objectives. Overall, we can be assured that the global network we have proposed can soon be augmented by systems of voice input and prose structure, leading to a new level of significance in information.
The other general realm of ‘artificial intelligence,’ or more accurately, approximations to organic action, by fully mechanized programs is equally appealing. Here, we have been applying for nearly a decade about 150 of what are now called expert systems, in operation of our own industry. These are spreading rapidly to other parts of manufacture and commercial activity, and even to areas like medical and hospital services. They have an important role in command and control for the military, where rudimentary forms were started years ago but only now are they being systematically applied. Obviously these operations support systems (as we named them before ‘support systems’ was stylish) have utterly central parts in crisis management and overall emergency reactions, although the appropriate implementation has been slow. What we are now urging is that the sophistication of these systems is such that they should form new integral parts of the global and national information arrays that we are discussing. These expert systems depend on clever programming of options for action which arise during an actual process or operation. Thus, they reflect evolving conditions which often simulate human judgment, and even intuitive interpretation of accumulating data on events, as they happen. Now however, we are seeking to provide such expert systems in new and compact form. This is arising as follows. Namely, a few years ago we determined to join the science of solids with a technology of integrated microcircuitry, thin film electronics, and software design, (although the latter is still grievously lacking in scientific principles). We did, however undertake to reverse the conventional and time-honored practice of designing an elegant electronic circuit for information processing and then adapting the program for the processing to that circuit. Rather, we now are able first to construct or at least to outline the program which is generally compatible with some generic computer operating system such as UNIX. We then design the chip or microprocessor that best executes that particular program. For we have learned enough, particularly with thin film epitaxy and the use of quantum well structures to get speedy and extensive (say a million components per chip) chips so that the program can be executed.
A present example of this was recently reported from our laboratory in which the inference engine and its knowledge base are combined on a piece of 2.5 micron CMOS patterned silicon. The first version of this circuit included only 16 rules. A next version, still having only 8,300 transistors, contained 64 rules, and the next one containing 256 cycles per inference yields 80,000 fuzzy logical inferences per second (FLIPS). The fuzzy element can cut to an eighth or more the throughput times by requiring only that fraction of rules that the logical inference per second system would use. The circuit design has tree-like sets of minimum and maximum detectors in parallel. Thus all the rules are evaluated simultaneously. Likewise, the rule storage on the chip has been changed from the usual read only memory (ROM) to random access memory (RAM) so the user can add acquisition of knowledge gained from ongoing experience. These tactics lead to useful output of the chip about 10,000 times faster than ordinary expert system operation on conventional computer.
It would seem that such capabilities could be applied in a host of demanding conditions for finding emergency measures in health, food supplies, water quality, military movements. Likewise, of course, the vital matters of information integrity, by cross checking and comparing large issues of information from any source would be enhanced. The mainstreams of crisis management programs will, in the Future Focus, in fact provide excellent examples and opportunities for application of these systems. Above all, however, is evident that expert systems with their versatility, speed and invulnerable objectivity can be major aids to integrity in a post-information age.
Although we have been able only to sample the range and diversity of ways of handling knowledge in this time of Future Focus, there has been consistent indication that human action remains unique. It remains separated by vast chasms of ignorance from the best we can do with digital or analogue machines and their networks. Nevertheless, in the future we should expect to find increasing information about the human information skills from the behavior of these rudimentary accessories to thought. The work in neurophysiology and cell biology is gaining in so far demonstrating the enormous elegance of nature's brains and controls. Clearly, we need especially to understand the interactions in groups of neurons, since their marvelous function does not seem to come from single synaptic processors, and the total phenomenology of the brain with the 10 13 or more neurons forming the basis for psycho science, is also far beyond our analytic skills. Hence, it is interesting that experiments by Lynn Jelinski, David Tank, and their associates in our laboratories, have sought to determine interactions among groups of neurons by observing their influence on thin film electronic circuits. H. J. Cheil in our Molecular Biophysics Research Department has been able to place neurons from the sea snail Aplysia Californica on such inorganic circuits. He then observes by interaction of the organic with the electronic inorganic substrate what some of the electrical behavior is like. Silicon nitride surfaces are particularly compatible, with adhesions of such invertebrates cells and conduction through the membranes also being observed. Martin Peckerar of the Naval Research Laboratories in collaboration with Dr. Richard Wyatt of St. Elizabeth's Hospital, recorded the electrical doings of up to 30 embryonic rat cells on the 30 electrode chip 3 to 5 microns wide, for activities lasting over about a week. They are hopeful that such probes can be implanted in living structures eventually.
Overall, we must expect continuing and exciting evidence of the information handling of cells and assemblies up to higher organisms themselves. We shall at least be challenged to keep improving our primitive approximations. But also in terms of trauma from disasters, emergencies, and conflicts, we shall learn something of the therapy in preservation of human nervous systems, which is in itself a major need in crisis management and survival.
The modern technology of communications and computers has many other meanings for our understanding of people and how their minds and bodies can adapt to change, and will respond to trauma of emergencies and difficulties, especially in the complex definitions of information-activated trauma that we mentioned at the outset. There is even significant progress in neuron modeling itself, beginning in our laboratories with the pioneering work of Harmon many years ago, when fact we learned something new about the spiral innervation of the cochleus. The study has moved into artificial neuron networks, in which NNOS amplifiers act as a body of neurons and interconnections, with variable resistors standing for the dendrites. The synaptic connection giving this variable resistance is difficult to put into modern integrated circuits. But our associates have reported a month or two ago that 2 layers of tungsten wires form a sandwich with a layer of amorphous silicon that provides a variable resistance connection. From this, a 512 neuron system with over 256,000 interconnections has been produced. It is being experimented with, in conjunction with a microprocessor for pattern and recognition. At the Jet Propulsion Laboratories other thin film techniques are yielding variable resistances, including those from a silicon hydride compound which is so versatile that perhaps a billion connections per square centimeter can be envisioned. At the Lincoln Laboratories at MIT, neurons are represented by charge packets from the charge coupled devices invented by Boyle and Smith at Bell Laboratories, and resistive connection are made from metal nitride oxide (MNOS) gates. These may, in fact, some day comprise programmable device assemblies, according to Dr. J. P. Sage.
Then there is a systems derivative of these modeling studies, particularly as generated at the Los Alamos National Laboratory called HUMTRN standing for Human Transport. This computer data facility provides simultaneous access to about 10 million pieces of information on the results of any chemically identifiable substance taken into the human body. This is reflected during activities of the body such as eating, breathing, working, perspiring, growing, and even dying. The inputs involve food, water, and air, etc. This model of assembled from the work of about 30 biologist and computer scientist, is a kind of appealing climax in concept to our theme for future focus.