Search

When writing the first draft of this book, some years ago, we had the idea to explain CogPrime by aligning its various structures and processes with the ones in the “standard architecture diagram” of the human mind.

Broadly speaking, the learning of predicates and schemata (executable procedures) is done in CogPrime via a number of different methods, including for example PLN inference and concept predicatization (to be discussed in later chapters). Most of these methods, however, merely extrapolate procedures directly from other procedures or concepts in the AtomSpace, in a local way—a new procedure is derived from a small number of other procedures or concepts.

Procedural knowledge must be learned, an often subtle and difficult process—but it must also be enacted. Procedure enaction is not as tricky a topic as procedure learning, but still is far from trivial, and involves the real-time interaction of procedures, during the course of execution, with other knowledge. In this brief chapter we explain how this process may be most naturally and flexibly carried out, in the context of CogPrime’s representation of procedures as programs (“Combo trees”).

“Language” is an important abstraction—but one should never forget that it’s an abstraction. Language evolved in the context of embodied action, and even the most abstract language is full of words and phrases referring to embodied experience. Even our mathematics is heavily based on our embodied experience—geometry is about space; calculus is about space and time; algebra is a sort of linguistic manipulation generalized from experience-oriented language, etc.

AGI design necessarily leads one into some rather abstract spaces—but being a human-like intelligence in the everyday world is a pretty concrete thing. If the CogPrime research program is successful, it will result not just in abstract ideas and equations, but rather in real AGI robots carrying out tasks in the world, and AGI agents in virtual worlds and online digital spaces conducting important business, doing science, entertaining and being entertained by us, and so forth.

This brief chapter deals with two important, coupled cognitive components of CogPrime: the component concerned with creating internal simulations of situations and episodes in the external physical world, and the one concerned with storing and retrieving memories of situations and episodes.

The artificial intelligence field is full of sophisticated mathematical models and equations, but most of these are highly specialized in nature—e.g. formalizations of particular logic systems, analyzes of the dynamics of specific sorts of neural nets, etc. On the other hand, a number of highly general models of intelligent systems also exist, including Hutter’s recent formalization of universal intelligence [Hut05] and a large body of work in the disciplines of systems science and cybernetics—but these have tended not to yield many specific lessons useful for engineering AGI systems, serving more as conceptual models in mathematical form.

In procedure learning as elsewhere in cognition, it’s not enough to use the right algorithm, one has to use it in the right way based on the data and context and affordances available.

Many different approaches to “low-level” perception and action processing are possible within the overall CogPrime framework. We discussed several in the previous chapter, all elaborations of the general hierarchical pattern recognition approach.

When one is involved with engineering an AGI system, one thinks a lot about the aspects of the system one is explicitly building—that are the parts, how they fit together, how to test they’re properly working, and so forth.

The primary burden of this book is to explain the CogPrime architecture for AGI—the broad outline of the design, the main dynamics it’s intended to display once complete, and the reasons why we believe it will be capable of leading to general intelligence at the human level and beyond.

In the previous chapter we presented a roadmap aimed at taking AGI systems to human-level intelligence. But we also emphasized that the human level is not necessarily the upper limit. Indeed, it would be surprising if human beings happened to represent the maximal level of general intelligence possible, even with respect to the environments in which humans evolved.

While we believe CogPrime is the most thorough attempt at an architecture for advanced AGI, to date, we certainly recognize there have been many valuable attempts in the past with similar aims; and we also have great respect for other AGI efforts occurring in parallel with CogPrime development, based on alternative, sometimes overlapping, theoretical presuppositions and practical choices. In most of this book we will ignore these other current and historical efforts except where they are directly useful for CogPrime—there are many literature reviews already published, and this is a research treatise not a textbook. In this chapter, however, we will break from this pattern and give a rough high-level overview of the various AGI architectures at play in the field today. The overview definitely has a bias toward other work with some direct relevance to CogPrime, but not an overwhelming bias; we also discuss a number of approaches that are unrelated to, and even in some cases conceptually orthogonal to, our own.

A program evolution component is proposed for integrative artificial general intelligence. The system’s deployment is intended to be comparable, on Marr’s level of computational theory, to evolutionary mechanisms in human thought. The challenges of program evolution are described, along with the requirements for a program evolution system to be competent- solving hard problems quickly, accurately, and reliably. Meta-optimizing semantic evolutionary search (MOSES) is proposed to fulfill these requirements.

While it’s widely accepted that human beings carry out both symbolic and subsymbolic processing, as integral parts of their general intelligence, the precise definition of “symbolic” versus “subsymbolic” is a subtle issue, which different AI researchers will approach in different ways depending on their differing overall perspectives on AI.

One of the most powerful metaphors we’ve found for understanding minds is to view them as networks—i.e. collections of interrelated, interconnected elements. The view of mind as network is implicit in the patternist philosophy, because every pattern can be viewed as a pattern in something, or a pattern of arrangement of something—thus a pattern is always viewable as a relation between two or more things. A collection of patterns is thus a pattern-network. Knowledge of all kinds may be given network representations; and cognitive processes may be represented as networks also; for instance via representing them as programs, which may be represented as trees or graphs in various standard ways. The emergent patterns arising in an intelligence as it develops may be viewed as a pattern network in themselves; and the relations between an embodied mind and its physical and social environment may be viewed in terms of ecological and social networks.

Having chosen to represent procedures as programs, explicit procedure learning then becomes a matter of automated program learning.

Most commonly, if a work on advanced AI mentions ethics at all, it occurs in a final summary chapter, discussing in broad terms some of the possible implications of the technical ideas presented beforehand.

A significant portion of CogPrime’s dynamics is explicitly goal-driven—that is, based on trying (inasmuch as possible within the available resources) to figure out which actions will best help the system achieve its goals, given the current context.

In this chapter and the next we describe one particular incarnation of the above ideas on semantic CSDLNs in more depth: the integration of CogPrime with the DeSTIN architecture reviewed in Chap. 5 of Part 1.

Creating AGI, we have said, is not only about having the right structural and dynamical possibilities implemented in the initial version of one’s system—but also about the environment and embodiment that one’s system is associated with, and the match between the system’s internals and these externals. Another key aspect is the long-term time-course of the system’s evolution over time, both in its internals and its external interaction—i.e., what is known as development.

The subtlest and most difficult aspect of logical inference is not the logical rule-set nor the management of uncertainty, but the control of inference: the choice of which inference steps to take, in what order, in which contexts. Without effective inference control control methods, logical inference is an unscalable and infeasible approach to learning declarative knowledge. One of the key ideas underlying the CogPrime design is that inference control cannot effectively be handled by looking at logic alone. Instead, effective inference control must arise from the intersection between logical methods and other cognitive processes. In this chapter we describe some of the general principles used for inference control in the CogPrime design.

Having discussed declarative, attentional, intentional and procedural knowledge, we are left only with sensorimotor and episodic knowledge to complete our treatment of the basic CogPrime “cognitive cycle” via which a CogPrime system can interact with an environment and seek to achieve its goals therein.

The approach taken to NLP in the OpenCog project up through 2013, in practice, has involved engineering and integrating rule-based NLP systems as “scaffolding”, with a view toward later replacing the rule content with alternative content learned via an OpenCog system’s experience.

Explicit knowledge is easy to write about and talk about; implicit knowledge is equally important, but tends to get less attention in discussions of AI and psychology, simply because we don’t have as good a vocabulary for describing it, nor as good a collection of methods for measuring it. One way to deal with this problem is to describe implicit knowledge using language and methods typically reserved for explicit knowledge. This might seem intrinsically non-workable, but we argue that it actually makes a lot of sense. The same sort of networks that a system like CogPrime uses to represent knowledge explicitly, can also be used to represent the emergent knowledge that implicitly exists in an intelligent system’s complex structures and dynamics.

Language evolved for dialogue—not for reading, writing or speechifying. So it’s natural that dialogue is broadly considered a critical aspect of humanlike AGI—even to the extent that (for better or for worse) the conversational “Turing Test” is the standard test of human-level AGI.

Language generation, unsurprisingly, shares most of the key features of language comprehension discussed in Chap. 26—after all, the division between generation and comprehension is to some extent an artificial convention, and the two functions are intimately bound up both in the human mind and in the CogPrime architecture.

Just as there are many different approaches to human flight—airplanes, helicopters, balloons, spacecraft, and doubtless many methods no person has thought of yet—similarly, there are likely many different approaches to advanced artificial general intelligence.

By way of conclusion, we now return to the “key claims” that were listed at the end of Chap. 3 of Part 1. Quite simply, this is a list of claims such that—roughly speaking—if the reader accepts these claims, they should accept that the CogPrime approach to AGI is a viable one. On the other hand if the reader rejects one or more of these claims, they may well find one or more aspects of CogPrime unacceptable for some related reason. In Chap. 3 of Part. 1 we merely listed these claims; here we briefly discuss each one in the context of the intervening chapters, giving each one its own section or subsection.

CogPrime’s knowledge representation must be considered on two levels: implicit and explicit. This chapter considers mainly explicit knowledge representation, with a focus on representation of declarative knowledge.

Spatiotemporal reasoning is an important skill that an AGI is expected to have, innately or not. Much work has already been done in defining reasoning systems for space, time and spacetime, such as the Region Connection Calculus for space, Allen’s Interval Algebra for time, or the Qualitative Trajectory Calculus for motion. However, these reasoning systems rarely take adequate account of uncertainty, which poses an obstacle to using them in an AGI system confronted with an uncertain reality. In this paper we show how to use PLN (Probabilistic Logic Networks) to represent spatiotemporal knowledge and reasoning, via incorporating existing spatiotemporal calculi, and considering a novel extension of standard PLN truth values inspired by \(\mathcal {P}(\mathcal {Z})\)-logic. This “PLN-ization” of existing spatiotemporal calculi, we suggest, constitutes an approach to spatiotemporal inference suitable for use in AGI systems that incorporate logic-based components.

“Procedure learning”—learning step-by-step procedures for carrying out internal or external operations—is a highly critical aspect of general intelligence, and is carried out in CogPrime via a complex combination of methods.

Intelligence is not just about what happens inside a system, but also about what happens outside that system, and how the system interacts with its environment. Real-world general intelligence is about intelligence relative to some particular class of environments, and human-like general intelligence is about intelligence relative to the particular class of environments that humans evolved in (which in recent millennia has included environments humans have created using their intelligence). In Chap. 4, we reviewed some specific capabilities characterizing human-like general intelligence; to connect these with the general theory of general intelligence from the last few chapters, we need to explain what aspects of human-relevant environments correspond to these human-like intelligent capabilities. We begin with aspects of the environment related to communication, which turn out to tie in closely with cognitive synergy. Then we turn to physical aspects of the environment, which we suspect also connect closely with various human cognitive capabilities. Finally we turn to physical aspects of the human body and their relevance to the human mind. In the following chapter we present a deeper, more abstract theoretical framework encompassing these ideas.

One of the hallmarks of general intelligence is its capability to deal with novelty in its environment and/or goal-set. And dealing with novelty intrinsically requires creating novelty. It’s impossible to efficiently handle new situations without creating new ideas appropriately. Thus, in any environment complex and dynamic enough to support human-like general intelligence (or any other kind of highly powerful general intelligence), the creation of novel ideas will be paramount. New idea creation takes place in OpenCog via a variety of methods—e.g. inside MOSES which creates new program trees, PLN which creates new logical relationships, ECAN which creates new associative relationships, etc. But there is also a role for explicit, purposeful creation of new Atoms representing new concepts, outside the scope of these other learning mechanisms.

Supposing the CogPrime approach to creating advanced AGI is workable—then what are the right practical steps to follow? The various structures and algorithms outlined in Part 2, should be engineered and software-tested, of course—but that’s only part of the study.

This chapter begins the heart of the book: the part that explains how the CogPrime design aims to implement roughly human-like general intelligence, at the human level and ultimately beyond.

In the last chapter we discussed human intelligence from a fairly down-to-earth perspective, looking at the particular intelligent functions that human beings carry out in their everyday lives. And we strongly feel this practical perspective is important: Without this concreteness, it’s too easy for AGI research to get distracted by appealing (or frightening) abstractions of various sorts.