I am currently thinking about creating a simple (but extensive extensible) mathematical system that would generate sentential statements (using numbers) about a symbolic concept. Some of the intermediate computations would signify a branching so the system would be able to generate an extensive number of values that branch out from the original computation. There are two important factors that are needed to make this work. One is a duplication of results that could be used to make the calculations compressed. A bucket kinds of things would work in this case because a collection of very different kinds of things could help disambiguate the meaning of the value (or symbol). So, I think this used to be called a hash table or something like that. However, I think the computational methods have to be designed to be a little more sophisticated than just using addition or multiplication. I am thinking of using the concepts that are used to determine how the computation takes place. (This is a very simple idea, I am talking about a numerical algorithm so it is just a subprogram.) However this creates a problem. Ordinary computation is efficient because it has a limited set of rules that are used consistently. I am thinking of an extensive set of rules that are determined by the concepts that are input (or the input values.) Since the system has to have an extensive reach, this means that there are going to have to be a lot of look-ups during a calculation. This may not be such a great problem because the computational rules are abstractions that may be kept separate from the rest of the data related to the conceptual knowledge related to some conceptual calculation. This may not be AGI but if some feasible compromise is possible, it might be used in an agi program. Jim Bromer
------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T395236743964cb4b-Ma8a1ef95320dc059d72da19c Delivery options: https://agi.topicbox.com/groups/agi/subscription
