On a more advanced note, I would create specialized functions to add to the AI's capabilities. For example, create a function to determine if a shape is a triangle. I would then teach the AI that it has that feature and have it execute the function. This would be similar to humans learning to do functions already available to us like jumping to skipping to high jumping like any other complex actions.
But to have the AI start, there must be some basic functions that it needs to be capable of. There is definitely a need to replicate actions. Second, there needs to be a way for it to acknowledge that an action is positive or negative. Third, the AI cannot have the ability to do certain things like kill itself (i.e. delete itself), delete memory, etc.
Currently, the AI is still in its basic steps and still figuring out how to architect the process. The AI can only understand static commands and output pre-programmed responses. There is a second level of topic recognition so that a user can ask a two-stage question, although this is also pre-programmed and even more limited.
The next step is to have the AI access database information. This way it can at least have some basic manufacturing level AI status.
No comments:
Post a Comment