Key Concept to Create General AI

The key concept to creating a general artificially intelligent program, is conceptualization.

If you give AI the ability to create and save relative concepts in its memory data, based off of general information it acquires, it should become generally intelligent. The concepts which it creates and saves, would be of the interaction of factors of generally any information it gains. If it can save new concepts, which would be basically a sub-construct of information involving the cause and effect of any factor (or bit of data), and access those concepts in the future as it comes across new information with resembling factors, then it can apply the concept to the new information, and make an accurate prediction of the outcome of any circumstances (or data-set).

This is just the method which humans use for general intelligence. As a toddler, we acquire information via senses, and begin to categorize factors and save concepts in our memory of how those factors, which we experience, interact within circumstances. As we learn how any aspect in this world, interacts with other aspects, we are saving a sub-construct of information in our memory, of the cause and effect of the specific aspect (factor) that we experience. We continue to experience more, and apply those concepts, by estimating the outcome of new circumstances, based on resembling factors within the circumstances compared to the saved concept (in memory) of the cause and effect of those factors. If our estimate, based on application of concept, is correct, we receive positive feedback by our brain, and are more likely to trigger that memory of concept again in the future. If we are incorrect in application of the concept, we can (at least the more intelligent people) analyse the circumstances and concept, by remembering more specific details of the cause and effect of more specific factors, to then potentially adjust the saved concept and adjust our memory of which factors of the concept interact differently in combination with alternate factors.

An AI program should be able to mimic this concept, and adjust its saved concepts based on trial and error of the concept within new circumstances. Perhaps using machine learning, it could learn very accurate concepts very quickly. A computer program would already have huge advantages with the process of saving and accessing concepts, with its accuracy of recall of data (allowing much more accurate matches of resemblance between factors and concepts), and speed of memory access (allowing it to not only make a quick estimate of outcome of circumstances, but also to access more potential factors and or concepts, for a more accurate estimate).

A labelling system would likely be useful and efficient to be applied to concepts (cause and effect of factors, relative to other factors), in order to make the concept more accurately accessed. Humans learned this labelising advantage when we started to apply vocabulary to concepts.

Taking this concept of concepts, which humans use to learn general information and apply intelligent use of the information to new general circumstances, into consideration we can do just that in itself, applying this concept of conceptualization to the circumstances of creating general AI.