The relation between semiotics and autonomy in intelligent systems is bidirectional: on the one hand, a system uses its shared and individual semiotics to maintain autonomy; and on the other hand, a system requires its autonomy in order to gather and maintain the information that is the basis for the formation of its semiotics. The feedback dynamics between semiotics and autonomy is very subtle and can be expected to manifest phenomena of chaos, complexity and emergence.
As autonomous intelligences, we experience this feedback from the inside; and as observers of other humans, we experience it from a combined internal/external perspective, due to our empathy for other humans and our embeddedness in the shared social semiotic dynamic of human culture and society. In engineering artificial intelligences, however, we get a unique opportunity to observe the semiotics/autonomy feedback from the outside, a perspective that yields many new insights. In this article I will describe Webmind, an Internet AI system which I have been developing for the past year, and I will discuss the relation between semiotics and autonomy in this system as it has evolved.
The main conclusion reached is as follows. In Webmind, semiotics supports autonomy via the social creation of self; and autonomy supports semiotics by supplying a coherent body of information to serve as raw material for the evolution of symbol groundings.
Webmind (Goertzel, 1997a), an intranet and internet-based AI system currently under development at IntelliGenesis Corp., embodies an understanding of intelligence as self-organizing, asynchronously distributed, and emergent, and provides one concrete vision of how genuine autonomous intelligence could be made to emerge from existing hardware and software.
From the user's perspective, Webmind is a general system for posing and answering questions regarding digitally stored information. It deals, potentially, with information of any kind, although, just as humans require eyes to perceive sights and ears to perceive sounds, Webmind must be given appropriate "perceptual methods" for processing each type of information into its own internal data structures. Its architecture is that of a massively parallel network, a population of many different static and dynamic agents continually recomputing their relationships with other agents, and acting on other agents in accordance with these relationships. The mix of different types of agents, and the amount of resources allocated to each, determines the emergent structure of the internal network, and hence the intelligence and functionality of the system.
Underlying Webmind is a vision of intelligence as "the capability to achieve complex goals in complex environments," which has been expressed mathematically in (Goertzel, 1993). The complex goals here are posed to Webmind by human users and evolved by itself, and the complex environments are the various parts of the human race and the Internet with which Webmind will have to interact.
Webmind is implemented in Java to run efficiently on powerful stand-alone computers, and in an ideal world would be run best on a supercomputer with multiple processors and tremendous amounts of random-access memory In the context of contemporary computing hardware, however, it is most cost-effective to run Webmind over a network of computers, in which case its sophisticated server-server communication methods allow its internal network structure to harmonize with the connectivity structure of the computer network.
The essence of Webmind's intelligence resides in the a portion of its code called the "Psynet." A Psynet is a self-organizing network of information-carrying agents. Information is incorporated into the Psynet via the creation of agents embodying that information. The architecture of a Psynet is relatively simple because the intelligence of the system is allowed to emerge from distributed interactions amongst the population of agents, rather than being imposed by specific reasoning rules or knowledge representation structures. The Psynet represents the minimum of structure required to lead to the adaptive emergence of useful information structures embodying data items. In short, the data stored in the Psynet is allowed to discover its own structure, within given constraints, rather than having structure imposed on it by rigid, preconceived rules. The design of the Psynet package is based on a mathematical theory called the "psynet model of mind," developed in a series of four books and numerous research papers over the period 1993-97 (see Goertzel, 1993, 1993a, 1994, 1997).
Agents within the Psynet are of three types: static, relational and mobile. Static agents may represent temporal data, but are static in the sense that they have a continued existence, maintained by the Psynet itself. Relational agents are not known directly to the Psynet but are held by other agents, representing relations between that agent and other agents. Mobile agents are like relational agents, but change frequently with time; they represent the learning of relationship by the Psynet's static agents. The Psynet supports many different types of static agents, tailored for particular purposes.
Static agents are also called "nodes," whereas relational agents are also called "links," a terminology which connects the internal structure of the Psynet with the external structure of the Internet and intranets in many useful ways. However, this language should not distract one from the fact that static and relational agents are much more substantial than the nodes and links found in some other artificial intelligence architectures, e.g. neural networks. A node within the Psynet is nothing like an individual neuron in the human brain, but might be more fairly compared with a neuronal group within the brain (consisting of 10,000-100,000 neurons tightly connected and oriented toward a single purpose). Psynet nodes cover a wide range of scales, from individual words to entire texts, data files and database records, categories of text, categories of words, trends of change over time in collections of data or collections of nodes, etc. Most abstractly, there are nodes corresponding to other Psynets with which there is interaction, and nodes modelling aspects of the Psynet itself, for purposes of adaptation and self-improvement.
The construction of nodes which refer to collections of other nodes is of particular importance. These nodes are called "concepts," and they provide Webmind's internal network with an hierarchical structure, complementing its primary associative structure. The superposition of hierarchical and associative structure is called a "dual network structure" and is essential for the emergence of intelligent activity and link patterns.
Learning in the Psynet takes place in five ways:
A query into the Psynet results in the creation of a new node, a "query node" which creates new mobile agents, which travel about within the Psynet and create new relationships with the query node. The answer to a query is given in terms of the relationships found by this agent-swarming process. The Psynet's introspection process involves continually querying itself, using queries based on queries it has been posed in the past, and particularly queries on which it has performed badly: in this way, it continually produces new knowledge in the areas in which it has proved deficient, and fills in gaps in its performance.
The final and in some ways most interesting part of the Psynet is its mechanisms for server-server interaction. An individual Psynet is, potentially, an autonomous Webmind. In practice, however, greater intelligence will be achieved by networking Psynets together in various ways. A collection of Webmind servers that interact intimately together, becoming a network-based rather than machine-based Psynet, is called a "Webmind unit." Elements of a Webmind unit are less like humans participating in a society, than they are like different lobes or hemispheres within a single brain. On the other hand, Webmind servers belonging to different organizations will generally be able to interact with each other only via the first two methods, or via the first method alone. There is a gradation between "social" and "intra-brain" interaction here, as opposed to the rigid division between individual and society that we experience as humans.
Finally, the social network of a Psynet plays an important role in guiding its introspections. A Psynet thinks about -- queries itself about -- those topics that it judges to be most important at present, as judged by several criteria: trends it has recognized in itself, trends it has recognized in its social group, and trends in what its users and its peer Psynets have identified as its deficiencies. The degree to which a Psynet pays attention to the opinions of another Psynet is determined in an intelligent manner, based on its experience with that Psynet and other Psynets' opinions, according to an algorithm drawn by mathematical models of human social interaction.
Webmind will be used to solve many problems that are fairly self-contained, detached from the flow and organization of human affairs -- "Find me information about crazed Third-world dictators; "What do the trends in Japan say about the U.S. stock market?", etc. Things become yet more interesting, however, when one envisions the same sorts of questions being asked regularly within an organization, about processes and structures within that organization. Instead of the stock market, one may have productivity statistics from various divisions of a company; and instead of newspaper articles, one may have reports generated within the company, e-mails sent within the company, etc. If Webmind is installed on the company's intranet, then real-time queries regarding relationships between textual, numerical and other data to do with the enterprise may be posed by any employee with computer access at any time. The result is that Webmind's intelligence is integrated with the social intelligence of the organization, and the individual intelligence of the employees.
Furthermore, the social dynamics of the different Psynets residing in different parts of an organization's intranet will grow to reflect the social dynamics of the individuals using those parts of the intranet. For instance, each Psynet will respond most effectively and rapidly to queries involving information which it stores locally; but the information that a certain Psynet within a Webmind unit stores locally may change over time, depending on user needs and internal Psynet dynamics. Thus, while providing easy access by all users to all information at all times, Webmind will nevertheless nudge the information at the readiest disposal of individual humans and divisions in certain directions, based on its inferences and its own emergent understanding. Webmind will do more than just provide an understanding of structures and processes; it will be a participant in processes, in the formation of emergent human and informational structures.
One consequence of this "artificial sociality" is that an individual Webmind's intelligence is crucially dependent on its understanding of itself in relation to other Webminds. It must know "who it is" in order to know what queries to handle internally and what ones to pass on to other Webminds. Thus, "autonomy" in the psychological sense -- the formation of a healthy self system -- is crucial for Webmind's intelligence. Without sociality and autonomy, the network of relations in an individual Webmind will not be adequately intelligent, because it will not draw effectively on the emergent knowledge of the community of Webminds
Next, to rigorously explore the question of semiotics in regard to Webmind, one requires a precise definition of "meaning." The definition to be used here is the one supplied in Chapter 5 of (Goertzel, 1994), where it is suggested that themeaning of an entity is simply the set of all patterns related to its occurence.For instance, the meaning of the concept "cat" is the set of all patterns, the occurence of which issomehow related to the occurence of a cat. Examples would be: the appearance of a dead bird, alitter box, a kitten, a barking dog, a strip dancer in a pussycat outfit, a cartoon cat on TV, a tiger, a tail,....
It is clear that some things are related to "cat" more strongly than others. Thus the meaning of "cat" is not an ordinary set but a fuzzy set. A meaning is a fuzzy set of patterns. In this view, the meaning of even a simple entity is a very complex construct. In fact, as is shown in (Goertzel, 1994), meaning is in general uncomputable in the sense of Godel's Theorem. But this does not mean that we cannot approximate meanings, and work with these approximations just as we do other collections of patterns.
This approach to meaning is very easily situated, in the sense of situation semantics (Barwise and Perry, 1981). The meaning of an entity in a given situation is the set of all patterns in that situation which are related to that entity. The meaning of W in situation S will be called the S-meaning of W. The degree to which a certain pattern belongs to the S-meaning of W depends on two things: how intensely the pattern is present in S, and how related the pattern is to W.
These ideas are not difficult to formalize. Consider a pattern in S
as a process p whose result rp is similar to S.
Let s be a "simplicity function" mapping the union of the space of mental
processes and the
space of situations into the nonnegative real numbers; and let d be a metric on the space of
mental processes, scaled so that d(rp,S)/s(S) 1/c represents an unacceptably large degree of similarity.
Then one reasonable definition of the intensity with which p is a patern
in S is given by the formula
Next, let MW,S(p) denote the degree to which p is an
element of the S-meaning of W (S is a situation, W is an object). Then one might, for instance, set
Finally, the relationship between meaning and intelligence is
worthy of brief comment. Intelligence, I have said above, is
considered as the capability to achieve complex goals in complex
environments. A mathematical definition of "complexity" may be
founded on the basis of the definition of "pattern" given above, the
gist of which is that an entity is complex if it has a large number of
different patterns in it. An environment is complex if it has many
patterns in it; and a goal is complex if, when considered as a
function mapping situations to degrees of goal achievement, its graph
has a large number of different patterns in it. The collection of
patterns in an environment or goal, which makes that environment or
goal complex, is a fuzzy set of patterns. Since intelligence has to
do with processing and enacting complexity, it requires manipulating
fuzzy sets of patterns. Meanings of specific elements in the world
may thus be viewed as subsets of the overall field of complexity in
which intelligence operates. Isolating meanings is a way of breaking
down complexity into parts -- i.e., it is an example of problem
solving by division and reunification, the most basic problem-solving
strategy of all.
A consequence of the above-given definition of meaning is that, in order for an intelligence to understand the meaning of a word, concept or other entity, this entity must be "grounded" in the system's own experience. Consider, for example, the position of Webmind or another similar intelligent text processing system on encountering the word "two." In order to deal with "two" in an intelligent way, the system must somehow understand that "two" is not only a pattern intexts, it also refers to the two computer users who are currently logged onto it, the two kinds of data file that it reads in, etc.
Through processing of text alone, part of the fuzzy set that is the meaning of "two" can be acquired: that part which is purely linguistic, purely concerned with the appearance of "two" in linguistic combinations with other words. But the other part of the meaning of "two" can only be obtained by seeing "two" used, and using "two," in various situations, and observing the patterns in these situations that are correlated with correct use of "two." In general, the fuzzy set of patterns that is the meaning of a symbol X involves not only patterns emergent among X and other symbols, but also patterns emergent among X and entities involved in physical situations. Recognition of patterns of this type is called symbol grounding (Harnad, 1990), and is a crucial aspect of artificial intelligence. Symbol grounding is, in essence, the difference between semiotics and formal language.
One approach to the task of symbol grounding would be to encode thousands or millions of specific rules, explaining which items in which situation s correlate with which symbols. However, this kind of approach is obviously doomed to fail, because even if one could accomplish such a tremendous feat of rule encoding for all known situations, one would still face the problem of enabling one's system to deal with unknown situations. Fortunately, in a self-organizing intelligent system such as WebMind, it is possible to take a different approach: to supply a small number of symbol-reality mapping rules and allow the remainder to emerge spontaneously from the network's own dynamics. The general principle here is that the subjective reality of an intelligent system is necessarily self-organizing: only a subjective reality which "understands its own structure" will be able to adapt and grow, and adaptation and growth are prerequisites of intelligence.
In Webmind, each static, relational or mobile agent has the ability to look inside itself and report various aspects of its current state. Textual statements, once read into the system, are decoded into collections of agents. A "symbol grounding" is represented by a special kind of mobile agent, a symbol grounding agent. A symbol grounding agent is associated with another agent, which it "tracks." Its goal is to mimic the behavior of the agent which it tracks, but without looking into that agent itself, only by looking into other agents. In this way it represents the agent that it tracks as a relationship among other agents. Given the Psynet itself as a fixed information base, the symbol grounding agent attempts to become a pattern in the node that it tracks. A given agent may be tracked by a number of symbol grounding agents, each of which is operative in certain situations, and some of which are more effective than others. The idea of meaning as a fuzzy set of patterns is thus embodied concretely: the meaning of an agent is enhanced by a fuzzy set of meaning-gathering agents. Situation specificity of meaning-contributing patterns is reflected as situation specificity of agents. These are intelligent agents, as recognizing patterns in the occurence of linguistic symbols or other mental entities is not an easy task; but the machine learning problems involved here are not insurmountable, a nd are in fact similar to those encountered in other aspects of Webmind, e.g. financial prediction.
The implementation of symbol grounding agents involves many detailed and difficult issues not touched on here, but the conceptual aspects of the design should be clear. The point is that the population of symbol grounding agents associated with a piece of information expresses that piece of information, explicitly, as a fuzzy set of patterns. It does so using the context of the Psynet. The usefulness of the symbol grounding agents for generating meaning depends on two factors: the intelligence of the agents, and the coherence and informativeness of the information in the Psynet itself.
From this detailed discussion, we may extract a few general points about meaning and mind:
In an autonomous intelligent system, such as Webmind or the human brain, each entity is a symbol for a tremendously large fuzzy set of other entities in the system. Meaning thus emerges as a self-organizing web of fuzzy patterns. The task of building a subjective reality is carried out by this web; and the subject ive reality constructed by this web is a key contributor to the web's growth. The Webmind software system provides a concrete, non-human context in which to examine these processes.
And how, then, is the feedback between semiotics and autonomy manifested in Webmind? Because the evolution of symbol groundings is derived by usage of the nexus of relationships in the Psynet, semiotics depends on autonomy, inasmuch as autonomy is what allows an integral, useful nexus of relationships to evolve in the Psynet. On the other hand, Webmind's autonomy relies equally much on its semiotics, in the sense that Webmind's understanding of itself hinges on its understanding of its relationship with humans and other computers, and its understanding of this relationship would be feeble indeed without symbol groundings of basic relational concepts. Essentially, we conclude that, in Webmind,