If we are to build artificial intelligent devices which will work along side humans and mimic much of the same methods of learning and thinking we will need to design a better data dumping system. Why? Well because as a computer system with artificial intelligence in the future will need to program itself thru its own observations, but as we all know sometimes when learning we observe something and interpret that data and then later make corrections. When learning a new skill or bettering our judgment in decision making we are completely re-adjusting and for artificial intelligence to do that also it must be able to continue its learning process. This is why we must include "data dumping” as one of the key features and one of the most important features in the development of self-learning, self-programming and next generation artificial intelligent computers.

But how can we make sure that this is done the most efficiently, after all if you dump the wrong data then you could be in big trouble, especially if the artificial intelligent android robot is making your dinner and burns up the kitchen. For instance if the meal is not perfect you do not want it to dump the entire recipe, only the part that was over or under cooked. Ideally the artificial intelligent robot could like your mother and grandmother adjust the recipe each time until every one being served is ultimately delighted.

Additionally it is important to "trash can” the information, but be able to retrieve it if needed sometime in the future. Or to dump partial data sets or replace them, but as it learns and experiments it will need to hold some of the old data, as it might be important data for future renditions of your recipe. So, please be thinking on the "data dump” concept if you are programming artificial intelligence and consider all this in 2006.