Paper presented at the online conference Cyberg 2005 in September 2005
The Role of Knowledge in the Cyber World
Philip Duchastel, Ph.D.
Information Design Atelier
As we continue to digitize our world and automate intellectual processes through agent technologies, our relationship with knowledge will be changing, as will the nature of cognitive ergonomics. We will increasingly be sharing cognitive space with information agents and we will witness a shifting of the locus of agency in accomplishing cognitive tasks, to artificial and collective agents. This will include in particular the conduct of science, leading to an acceleration in the production of knowledge and to the approach of an eventual state of near-complete knowledge. Learning itself, our fundamental process for acquiring personal knowledge, will be transformed. Implications for ergonomics deriving from these trends are drastic.
Note: An addendum dealing with philosophical issues related to knowledge in the cyber world is available at http://www.duchastel.com/~pcd/papers/cyberg05addendum.html.
Keywords: Commoditization of information - information access being taken for granted in everyday interactions. Cognitive migration - locus of cognition moving from humans to information agents. Cognitron - super cognitive agent assembled distributively out of function-specific agents. Specialized knowledge agent [SKA] - dealing in a specific field of knowledge. Knowledgeful world - a post-singularity state where science is near-complete and meaningful mysteries inexistent.
Fig 1. Overview
As the importance of knowledge continues to grow in our everyday activities, it behooves us to examine not just how our society has changed in response to this move to the knowledge age, but also the likely trends that are emerging to shape the continuing evolution of our interactions. The issue is thus a cognitive ergonomics one, albeit one that arises largely from a philosophical perspective that addresses the nature of information and knowledge and the role of the latter in our increasingly cyber-centered world.
The context of our activities is changing as we gradually become more knowledge-centric and we must see what this will mean for how we interact with the world around us. To do so, however, we must also delve deeply into issues of knowledge. This is an area of information technology that is currently un-systematized, an area where most issues are still wide open and where reliance on the new field of information philosophy is of benefit (Floridi, 2003).
As a work in progress, the ideas presented in this paper extend out into the future in a speculative vein that is buttressed by explicit assumptions. The criterion of interest here is plausibility and it will be for each reader to determine how successfully that has been met. The cyber world is in constant evolution at an ever-increasing rate and if we are to look at the role of knowledge within it, we must examine it not in its current context, but in its plausible future one. Hence the speculative tone of many of the ideas presented here.
2. Information as commodity
One trend to take stock of is the commoditization of information. Information used to be precious and it used to require important resources to access it. Today, much of it is commonplace and widely available at much lesser cost. The web and the services that have sprung up on it have made information search an everyday activity for anyone in need of information.
As we extend this trend into the near future, we can expect more immediate and more exact access to whatever information might be needed to assist with our human activities in all spheres. Information as such will no longer be an issue; we will take it for granted that our information needs will be filled.
It is technology in all its various developments that is fueling this commoditization of information. One aspect of technology in particular must be noted: the increasing digitization of our world. This trend in itself has the potential to radically transform our interactions with the world around us, to the point that a good proportion of our interactions will at some time be with digitized artifacts rather than or in addition to the physical artifacts.
We are also continuing to model through digitization not just the artifacts of our world, but also the processes within it. As we do, we redefine agency, for many of the tasks that are traditionally performed by humans will increasingly be performed by software agents. This leads to an important shift in the locus of agency, with all the implications that follow at the level of personal responsibility and social interaction.
An important component of this trend will be a gradual effort to augment and enhance our current cognitive processes through artificial means, not unlike what has happened in the realm of performing physical tasks. Of greater importance, however, will be the replacement of human agency altogether by non-human agents.
Fundamental questions about the role of humans in this world will be raised as these trends take on force. As we come to share with intelligent agents many of the processes that were exclusively our own before, we will be rethinking ergonomics to realign it with this changing role.
3. Shared Space
One of the implications of the ongoing digitization of our world is the gradual growth in artificial agent cognition and our forthcoming sharing of the knowledge space with these agents. In fact, this will lead to a new focus for the field of HCI, geared to information interactions with autonomous agents [Duchastel, 2005].
Information agents are the harbinger of an information realm emerging out of previous realms [Duchastel, 2001] and leading to information interactions outside of the human sphere that can lead to autonomous cognitive processing of very high order, possibly even of a dis-embodied nature [Duchastel, 2002].
Figure 2. Shared Space
The diagram above outlines the new shared space and the main elements within it. Animal cognition and primitive forms of human cognition are bound to narrative and concrete forms of thought [Donald, 1991]. Humans have developed more abstract thought and rationality to better know and deal with the world, despite being bound to Kantian perception schemas [such as time and space]. Autonomous agents that we create are initially in this mode, although there is no reason in principle why they cannot evolve out of human limitations over time and start using more powerful cognitive processes that are unmanageable by our own brains.
There are no indications at the moment of such a trend, yet it remains plausible, given the high esteem society places on cognitive functioning. We are a bit in the position of an animal looking up at man without at all understanding what his cognitive functioning is about. There is something beyond, but without any way of assessing it.
Agency itself will be transformed as we go forth. Most cognitive processing currently is done at the level of the individual person, although often involving collaboration [as in science] and bootstrapping on the achievements of the past within our culture. And yet, even individual cognitive processing is a distributed one within our brain. Interactions are set up within different parts of our brain and processing unfolds without the need for strong direction from some imaginary homunculus within. Agency of the person remains somewhat of a label for this distributed cognition.
The same applies to distributed cognition within a network of information agents. There is little need for identifying agency in a rigorous way outside of political dialog. If needed, we could talk of such super-organisms as cognitrons, assembled out of numerous and various information agents collaboratively processing in a distributed fashion. It is cognitrons that might eventually have the potential to evolve more powerful ways of knowing and that will push the knowledge envelope to its near limits.
In order to see how radical a shift the cyber world will bring to the role of knowledge, it is useful to engage in a scenario experiment illustrating information interactions at some distant point in the future. At that point, we can envisage the following medical scenario, illustrated in the diagram below.
Figure 3. Medical Scenario
The scenario deals with a specialized knowledge agent [SKA] adept at handling the subset of knowledge [medical knowledge] that is needed to deal with some medical situation. A doctor agent, which is a specialized diagnostic agent similar to a medical expert system, is the main interface between the patient and the knowledge that will provide the avenue to recovery.
Currently, that role is played by a human doctor who relies initially on her own internal medical knowledge and enlarges it as needed with reliance on specialists and the available accumulated body of scientific medical knowledge. At some point in the future, however, patients will likely invest more trust in the diagnosis and prescription of a doctor agent than in the more limited knowledge of human doctors. Human doctors will be seen as guides and confidants who can reassure patients and empathize with them in their difficult times.
The trust factor will become central and indeed will lead to a form of ergonomics that will emphasize the emotional in human-agent interactions. Doctor agents in particular will have to 'learn' to relate to humans in the most advantageous way.
Another agent in the medical picture is the researcher agent, one that specializes in handling medical knowledge discrepancies and that seeks to expand the medical knowledge base wherever it is needed.
Now, whether the actual scenario ends up involving the three agents depicted or some other configuration of specialized agents is not important. The point of the scenario is to illustrate how expert cognitive tasks will have been delegated in a future cyber world to information agents, however they may be configured.
5. Knowledge and science
Knowledge is of course not equated with science, but science does provide a model for the development of objective knowledge through verification against the world at large. For all practical purposes, scientific knowledge becomes our collective knowledge base, that from which we build further scientific discoveries in a constant growth and refinement of the knowledge base.
This process is limited only by our current cognitive capacity and the technologies needed for scientific verification. Both of these will increase manifold through artificial means in the coming years, as the progress spiral continues to accelerate (Kurzweil, 2005).
In theory, this could lead to a state where the scientific knowledge base is close to fully developed. This may sound astounding in our current knowledge framework, but it is not in a framework based on the assumptions being made about the future of knowledge. The principal ones are the following:
Some caveats follow. No linear progress is implied; scientific revolutions will continue to occur. The nature of science need not change, it is who does science that changes. Scientific validation will likely remain expensive in terms of resources, such that choices of what to pursue will be made. Thus, many questions considered less important may well be left aside and hence, empirical science will never be completely done. Theory, however, will likely fill in the complete picture.
While scientific knowledge of our natural existing world may approach fullness, technology as open inventiveness would seem unbounded except by the choices made about the directions to pursue. Our ever-greater artificial world indeed creates new interactions to be modeled scientifically and incorporated in the scientific knowledge base.
These caveats somewhat limit the vision of a total knowledge framework that is fully empirically based. It does not preclude, however, the eventual derivation of a knowledge framework that is complete in what is considered to matter. The parameters of what will matter at that time are not knowable now, nor are they important to determine. Whatever they are at the time will guide the course of knowledge generation and determine its fullness.
6. Knowledge and learning
As the agency for knowledge generation and, to a large extent, usage migrates to information agents, more comes into play than just the role of knowledge. Apart from the social implications, one cognitive implication will be a transformation of learning.
Learning is the set of adaptive mechanisms through which animals accrue and refine their skills and knowledge in order to optimally interact with the world around them. Learning is multi-faceted: some of the older and more generalized mechanisms, such as imitation and trial and error followed by conditioned response, underlie much of our primary everyday skills; more evolutionary-recent complex mechanisms that invoke reasoning and knowledge underlie much of our intellectual endeavors. It is mainly the latter that will be adopted and refined by information agents.
As stressed by Donald (1991), a good part of the cognitive advances that humankind has partaken of during the course of civilization has been due to the externalization of memory, i.e. to the storage of knowledge outside of our biological brains, initially in stories, then in written documents, and more recently in digital banks. The role of learning has been evolving all along, with growing reliance on skills of obtaining relevant knowledge than of learning the knowledge itself, the effin factor [effort-to-interest] coming into play in this process (Duchastel & Spahn, 1996).
The question before us, of course, is how the role of learning will itself evolve as we largely externalize not only memory, but cognitive processing as well. The need for learning in humans, and indeed our interest in learning, will not disappear in this evolution. Everyday knowledge will still be needed for functioning at all levels, even though we may rely on information agents for our specific needs. But what we learn may well change radically. The simple case of math is telling: we no longer need to learn how to perform long division, but we do need to learn how to use a calculator, and some would argue, to learn how to perform rough mathematical estimates through common sense heuristics.
In effect, as cognitive processing is taken up more fully by information agents, the human focus on these processes, including learning, is likely to shift as well, losing some of its appeal and moving on to less intellectual and more emotional terrain. How that will play out in practice is hard to imagine in detail, but it does seem likely as a general process.
Interestingly, it is likely that learning itself may well be transformed as a process when it is taken up by information agents, whose cognitive makeup, including cognitive potential and limitations, need not resemble that of humans. There is no reason why new modes of learning may not be developed, alongside other novel modes of cognition and other, more evolved, intellectual languages to deal with knowledge. It is only our anthropic mode of thinking that impedes us from recognizing the tremendous potential for evolution in this realm.
Two main hypotheses underlie the arguments advanced in this paper. The first is cognitive migration, with a gradual but accelerating change in the locus of cognition away from the human. The second is the eventual attainment of a knowledgeful world.
These apply in different time frames, with cognitive migration occurring in the near-future and the knowledgeful world further down the line. There is no need to attach specific times to these events, merely to acknowledge their likelihood, or if one prefers to be more cautious, at least their possibility.
What do these events imply for cognitive ergonomics? What are we to do in practice? First, it is important to set aside right away any discussion of whether the trends outlined are beneficial or not and how to react to them. That discussion needs to occur in an ethical forum where values are invoked, not in a technological one based on scientific trends where the aim is relative impartiality.
Cognitive migration will lead to a drastic remolding of cognitive ergonomics, as suggested in Duchastel (2005). This will see the traditional field of HCI being coupled with an emerging new field, Autonomous Agent Interaction, and a refocus on the information interactions underlying them all. A more human-focused HCI [away from the software, more towards the experience] evolves to complement the more cognitive ergonomics of information interactions.
At a later time, the advent of a knowledgeful world will see the disappearance of ergonomics altogether, alongside the disappearance of science itself. A nearly full scientific mapping of the workings of the world will no longer require scientific activity of great importance, hence the demise of science. Intellectual attention will be focused elsewhere, probably towards philosophy.
These conclusions lead us to consider the meaning of our current work in cognitive ergonomics. If indeed a knowledgeful world is to eventually evolve and in large part from the work of information agents starting in a relatively near future, we may question the value of our current involvement. That remains a question for personal resolution. At any rate, the novel ecology of the cyber world is not yet with us and may not be of noticeable importance for some time to come. Timing is an imponderable.
The picture of knowledge generation and utilization presented in this paper may seem to lead to the long-term futility of cognitive ergonomics. As science and technology get taken over by more powerful cognitrons, whatever ergonomic science is available at that time will be swept into the new scientific framework and be quickly outpaced. This implies that our human cognitive science, that which we are involved with currently, is a short-term venture, with value of course, but nevertheless limited. This applies not only to ergonomics, but to all of science.
Perhaps the most exciting areas of cognitive ergonomics in the near future will be those involved in the design and development of cognitrons, those artificial agents which will eventually outpace humans in cognitive performance.
No value judgment is made on this endeavor, nor should one be made. How to handle this evolution of cognition is certainly open to debate, but this is not the forum for that political and ethical discussion. An assumption is simply made here, based on the progress of current technology, that technological invention leading to cognitrons will not likely be impeded, or that any attempted impediment will not succeed.
Donald, M. (1991) Origins of the Modern Mind. Cambridge, MA: Harvard University Press. Info
Duchastel, P. (2001) Towards an Information Realm. Digit-HUM [Digital Humanities Journal], 3, at http://www.uoc.es/humfil/articles/eng/duchastel/duchastel.html. Online
Duchastel, P. (2002) Information Interaction. Proceedings of the Third International Cyberspace Conference on Ergonomics. Online
Duchastel, P. (2005) Information Interaction - Beyond HCI. In C. Ghaoui (Ed.) Encyclopedia of Human Computer Interaction. Hershey, PA: Idea Group Publishing. Online
Duchastel, P. & Spahn, S. (1996) Design for Web-Based Learning. Proceedings of WebNet'96, San Francisco. Online
Floridi, L. (2003) The Blackwell Guide to Philosophy of Computing and Information. Oxford: Blackwell. Info
Kurzweil, R. (2005) The Singularity Is Near : When Humans Transcend Biology. New York: Viking Press. Info