Balancing Opportunities and Investments for NBIC

R. Stanley Williams and Philip J. Kuekes, Hewlett Packard Labs

Over the course of the last several millennia, human beings have learned that major tasks can be performed much more efficiently by dividing up the workload and sharing it among individuals and groups with specialized skills. Larger and more complex tasks require societies with more capable tools and communications skills. As we view the beginning of the 21st century, the tasks we want to perform have become so complex and the tools we have created so sophisticated, that we are challenged to even describe them coherently. It is time to take a holistic view of how we relate to our technologies and develop strategic approaches to integrating them in a fashion that makes them more adaptable and responsive to human desires and capabilities.

In 2001, we are seeing the simultaneous beginnings of three great technological and industrial revolutions that will spring from advances in fundamental research during the past two decades:

Information Science — the understanding of the physical basis of information and the application of this understanding to most efficiently gather, store, transmit, and process information.

Nanoscale Science — the understanding and control of matter on the nanometer length scale to enable qualitatively new materials, devices, and systems.

Molecular Biology — the understanding of the chemical basis of life and the ability to utilize that chemistry.

The knowledge base in each of these areas has the capacity to increase exponentially for several decades into the future, assuming that the research enterprise is maintained. Each field, by itself, offers tremendous opportunities and potential dangers for society, but the fact that there are three simultaneous technology revolutions is literally unprecedented in human history.

The greatest prospects and challenges will occur in the overlap areas that combine two or all three of the new technologies. The great difficulties are that (1) each area by itself is so large and intricate that no single human being can be an expert in all of it, and (2) that each area has developed a language and culture that is distinct and nearly incomprehensible to those working in the other areas. Thus, we find that the most significant problems are often not those related to any particular technology but are based on the basic inadequacies of human understanding and communication. This all-important human factor requires that we better understand and apply cognition. Cognitive science will become an increasingly important field for research and utilization in order to more effectively employ the technologies springing from information, nanoscience, and molecular biology. In turn, these technologies will enable major advances in the study and applications of cognition by allowing the construction and emulation of physical models of brain function.

A concrete example can help to illustrate the potential of these overlapping technologies. Since 1960, the efficiency of computing has increased approximately two orders of magnitude every decade. However, this fact has rarely been factored into solving a grand challenge by trading off computation for other types of work as an effort proceeded. This is largely because humans are used to maintaining a particular division of labor for at least a human generation. When paradigms change at a rate that is faster, humans have a difficult time adjusting to the situation. Thus, instead of a smooth adoption of technological improvements, there are often revolutionary changes in problem-solving techniques. When the human genome project began, the shotgun approach for gene sequencing was not employed, because the speed of computing was too slow and the cost was too high to make it a viable technique at that time. After a decade of steady progress utilizing, primarily, chemical analysis, advances in computation made it possible to sequence the genome in under two years utilizing a very different procedure. Thus, the optimum division of labor between chemical analysis and computation changed dramatically during the solution of the problem. In principle, that change could have been exploited to sequence the genome even faster and less expensively if the division of labor had been phased in over the duration of the effort.

As long as technologies progress at an exponential pace for a substantial period of time, those improvements should be factored into the solution of any grand challenge. This will mean that the division of labor will constantly change as the technologies evolve in order to solve problems in the most economical and timely fashion. For computation, the exponential trend of improvement will certainly continue for another ten years, and, depending on the pace of discovery in the nano- and information-sciences, it could continue for another four to five decades. Similar advances will occur in the areas of the storage, transmission, and display of information, as well as in the collection and processing of proteomic and other biological information. The route to the fastest solution to nearly any grand challenge may lie in a periodic (perhaps biannual) multivariate re-optimization of how to allocate the labor of a task among technologies that are changing exponentially during execution of the challenge.

These thrusts in twenty-first century science are being recognized by those in academia. Some university deans are calling them the "big O's": nano, bio, and info. These are seen as the truly hot areas where many university faculty in the sciences and engineering want to work. In looking further into the future, we believe that cogno should join the list of the big O's.

One way in which academe responds to new opportunities is by creating new disciplines at the intersections between the established divisions. Materials science was created early in the last century at the boundary between chemistry and structural engineering and has evolved as a separate and highly rigorous discipline. Computer science was created in the middle of the last century at the boundary of electrical engineering and mathematics. Now we are beginning to see new transdisciplinary groups coming together, such as chemists and computer scientists, to address new problems and opportunities. One of the problems we face at the turn of this century is that as device components in integrated circuits continue to shrink, they are becoming more difficult to control, and the factories required to build them are becoming extraordinarily expensive. The opportunity is that chemists can inexpensively manufacture components, i.e., molecules, very precisely at the nanometer scale and do so at an extremely low cost per component. Therefore, the new discipline of molecular electronics is arising out of the interactions between computer scientists and chemists. However, developing this new field requires the rigor of both disciplines, the ability to communicate successfully between them, and the proper negotiation process that allows them to optimally share the workload of building new computers. Chemists can make relatively simple structures out of molecules, but they necessarily contain some defects, whereas computer scientists require extremely complex networks that operate perfectly. Economic necessity brings these two very different fields together in what is essentially a negotiation process to find the globally optimal solution of building a working computer from nanometer scale objects at a competitive cost.

There are other very interesting examples of different sciences just beginning to leverage each other. In the bio-info arena, Eric Winfree at the California Institute of Technology is using DNA for self-assembly of complex structures by designing base-pair sequences to construct nano-scaffolding. There is also the whole area of the interaction between biology and information science known as bioinformatics. With the discovery and recording of the human genome and other genomes, we essentially have the machine language of life in front of us. In a sense, this is the instruction set of a big computer program that we do not otherwise understand: we have only the binary code, not the source code. There is a huge amount of work to reverse-engineer this binary code, and we are going to have to rely on computing power to understand what these programs are doing.

Another arena of extreme importance is the bio-nano intersection, since at the fundamental level these both deal with the same size scale. There will be tremendous opportunities to design and build measurement devices that can reach to the scale of molecules and give us a lot more knowledge about biology than we have now. But the reverse is also true. We are going to learn new ways to manipulate matter at the nanoscale from our huge investment in biology. The current goal of molecular electronics is to combine simple physical chemistry with computer design. But biomolecules have incredible functionality based on four billion years of R&D on very interesting nano-structures. The world is going to make a huge investment over the next few years in the biosciences, and we will be able to leverage much of that knowledge in engineering new nanoscale systems.

Work on the relationship between cognition and information goes back the Turing test (i.e., a test that determines if a computer can fool a human being into thinking it is a person during a short conversation) — ideas Turing had even before computers existed. As more powerful computers have become cheaper, we now have cars that talk to us. How will the next generation of people respond when all kinds of devices start talking to them semi-intelligently, and how will society start reacting to the "minds" of such devices? As well as the coming impact of info on cogno, we have already seen the impact of cogno on info. Marvin Minsky, in his Society of Mind, looked at the cognitive world and what we know about the brain and used that to work out a new model of computation.

With nanotechnology literally trillions of circuit elements will be interconnected. There is a set of ideas coming out of the cognitive science community involving connectionist computing, which only starts to make sense when you have such a huge number of elements working together. Because of nanotechnology, we will be able to start experimentally investigating these connectionist computing ideas. The other connection of nanotechnology with the cognitive sciences is that we will actually be able to have nonintrusive, noninvasive brain probes of conscious humans. We will be able to understand tremendously more about what is going on physically in the brains of conscious minds. This will be possible because of measuring at the nanoscale, and because quantum measurement capability will provide exquisitely accurate measurements of very subtle events. Over the next couple of decades, our empirical, brain-based understanding in the cognitive sciences is going to increase dramatically because of nanotechnology. The hardest challenge will be the bio-cogno connection. Ultimately, this will allow us to connect biology to what David Chalmers recognizes as the hard problem — the problem of the actual nature of consciousness.

The topic of discussion at this workshop is literally "How do we change the world?" What new can be accomplished by combining nanoscience, bioscience, information science, and cognitive science? Will that allow us to qualitatively change the way we think and do things in the 21st century? In the course of discussions leading up to this workshop, some of us identified nano, bio, and information sciences as being the key technologies that are already turning into 21st century industrial revolutions. Where do the cognitive sciences fit in? One of the major problems that we have in dealing with technology is that we do not know how we know. There is so much we do not understand about the nature of knowledge and, more importantly, about the nature of communication. Behind innovative technologies and industrial revolutions there is another dimension of human effort. In order to harness the new scientific results, integrate them, and turn them into beneficial technologies, we need to strengthen the cognitive sciences and begin the task of integrating the four big O's.

0 0

Post a comment