Toward Human-Level Cognitive Adequacy
Our long-range aim is to design and implement common sense in a
computer. Click here
for details.
If you would like to learn more about Active Logic, we suggest
you start with one of our primers.
An example of a misidentification error is recognizing an object (e.g., an enemy plane in restricted airspace) and drawing various inferences about it (it is going too fast to force down, it must be shot down) and then discovering it is not an enemy plane after all, and that the belief that it must be shot down should be rejected, whereas the belief that it is going too fast to force down should not be rejected (it still is violating restricted airspace and if it cannot be forced down then it still must be dealt with somehow but not fired upon).
We have found automated solutions to problems of this sort using active logics, where there is no “canned” solution as to what to do when a plane is first misidentified, conclusions are drawn, then the error is found. We have general inference rules that make use of the history and also of the Now predicate, that allow the system to figure out on its own what past conclusions it should reject and which to keep, on the basis of its knowledge about the world.