Machines that
“Understand” each Other
Research Professor Paul S. Prueitt
George Washington University
Information Technology has social roots as well as dependencies on computer technology and mathematics. These social roots should be reflected in our university and school curriculum. But the social aspect of information technology has been difficult for members of the Academia to see. As a consequence, the curriculum does not provide a common background that allows individuals to feel comfortable with the concept that a machine can “understand”.
In the past several decades, the term “information” is understood in a new sense. Data structures, and algorithms on these structures, are forming an interconnectivity called the semantic web. But the term “information” is also understood in a social sense. To what extent does social science and computer science curriculum agree or disagree on the meaning of the term “information”. This question is debated unequally within the Academia.
Some scholars claim that how we understand the term “information” has not been properly examined and when examined leads to surprises. These surprises are not consistent with how computer scientists have defined the term “information”. For example, scholars in the biological and social sciences have observed that the nature of computing is unlike the nature of living systems. Some logicians and mathematicians argue that a computer program, even one that is executing in a physical computer, is “mere” abstraction. Sir Roger Penrose, for example, uses difficult arguments from quantum physics and the foundations of logic and mathematics to make a distinction between the algorithm and natural processes. His is not the only argument that makes a similar point.
There is more that one side to this debate. Internet visionary, Tim Berners-Lee, speaks about there being “two sides” to the Internet, the machine consumable side and the human consumable side”. Other visionaries speak about a future ignition of the semantic web when the connectivity and interoperability of the machine consumable side reaches a certain threshold number, representing the ratio of connected systems to total systems. This notion of an ignition within a switching network follows the work of Stu Kauffman at the Santa Fe Institute. But why “two” sides. Is this two sidedness an explicit recognition that social science regards human reasoning and social knowledge as being non-machine like in nature?
We await interoperability between computer systems employing controlled vocabularies and machine-readable ontologies. In fact, the recent US/Iraq war used military ontology to provide a new level of war fighting capability. But many in the intelligence community are concerned about an observed impedance mismatch between what human analysts have been expecting from machine intelligence and what procurement processes deliver. The fact is that cultural resistance is based on a confusion that is not addressed by our educational system.
How might this mismatch be understood? One might turn to the life sciences to understand what might be incorrect about the information science. When this is done, one observes that a proper social science theory about the nature of information is hidden for two types of reasons.
The first type of reason arises from the difficulty of understanding the nature of human knowledge sharing and of cognitive dimensions to the exchange of social information. One needs a background in linguistics, in cognitive neuroscience, and in the social sciences. And the background needs to be the right background, because these disciplines are not perfect.
One also needs some type of formalism like mathematics or logic, and yet formal mathematics and logic are argued by a long line of scholars as being somehow limited in applicability to the social sciences and to biology. This scholarship includes works by J. S. Mill, C. S. Peirce, K. Godel, I. Rashensky, R. Rosen, D. Pospelov, R. Penrose, P. Kugler and others. Society does not as yet have a clear notion of how to cultivate science on knowledge sharing systems. But there is relevant scholarship available. So how does a student sort this out?
There is also a second type of hidden reasoning. This hidden reasoning arises from a category of social behavior related to fundamentalism. One can define what is meant by the term “fundamentalism” to mean any way of thinking that relies on a reduction of truth-finding to a set of primitives that are accepted to be true for purposes of setting issues of judgment. In this definition, the viewpoint of logical positivism shares membership within a category of viewpoints that includes also the religious fundamentalisms as well as the fundamentalism of secular humanism and scientific reductionism. Each of these has a consequence within the social discourse. Western philosophy sees this problem in terms of the “problem of other minds”. It is not only important how this problem is addressed, but also how it is framed.
Tim Berners-Lee’s notion assumes that informational interoperability can easily occur between social systems when the machine-readable ontologies allow machines to “understand each other”. But issues are raised by empirical observation of social reality. Fundamentalism can act to reinforce a specific viewpoint and to exclude other viewpoints. The consequence can be non-interoperability between social systems.
We return to the social theory and to our cultural need to understand this brave new world.