Goals, Decision Making, Problem-solving and Reasoning
In the early days of AI much of the work was around decision-making, reasoning and problem solving systems, often operating in a very abstract or block-world type environment. Many of the early cognitive architectures discussed in Chapter 6 started off in this form, such as SOAR and ACT-R.
The most common current model for generic decision-making in AI type systems is probably the Belief-Desire-Intentions (BDI) model (Rao, 1991), based on an evaluation of which plan meets the needs of the current goals and world-state.
Pomerol (1997) identifies a wider range of decision-making techniques that an AI could make use of including: utility functions, pattern matching, expert systems, numerical methods, neural networks, and case-based reasoning.
Another approach has been to try and give systems a level of common-sense to help in reasoning and decision making. Pease (2000) describes the US Defense Advanced Research Projects Agency (DARPA)’s High Performance Knowledge Bases (HPKB) project which focused on creating very large and fast knowledge bases for common-sense reasoning in the domains of warfare and geography. Another such example is the Open Mind Common Sense Project described by Singh (2012) .
The ultimate way to address these issues though will be through an Artificial General Intelligence, which may, or may not, incorporate some of these approaches.
World Models
In order to deal with the world(s) (physical or virtual) hypothetical situations and thought exercises the virtual human may need multiple world maps and models, one of which deals with the physical world (and perhaps others for different virtual worlds and other realms) and one or more for different hypothetical or imaged worlds. Such world models are likely to contain:
• Spatial relationships;
• Social relationships;
• Models of object interactions (e.g. the laws of physics);
• Models of social interactions; and
• Models of psychological behaviors.
The virtual human will also need some elements of a theory of mind (Goldman, 2012) so that it can also model what other people might know.
- Up to Mind
- On to Creativity and Imagination