True intelligence assessment:
- is only achievable through mass trials of test objects, with full control over their knowledge, with full reset capability of that knowledge and a total cultural knowledge when comparing the acts of the test objects with the optimal behavior according to the culture.
In essence this means that the assessment can only be performed on artificial test objects, such as UAV:s in aerial security culture, that the UAV experience of the trial can be erased, that the experimental environment can be restored properly, and that this environment can be subtly varied randomly so that chance lucky outcomes can be ruled out from affecting the assessment.
On the other hand, some degree of self awareness required to implement true intelligence can give some indication that real intelligence might be present in a test object:
- Graceful degradation: an unpassable obstacle in the road to fulfilling the plan ought to make the test object to give up and perform a default action, such as returning to the starting point and make a failure report,
- Damage free solution: on the road to plan fulfillment dangers must be minimized so that the test object avoids partial solutions that are self destructive, and that doesn't cause untolerable damage to objects in the test environment,
- Plan repair: if at carrying out a plan, the test object encounters an unexpected but solvable obstacle, the test object ought to clear away or circumvent that obstacle according to the test object's own prepared strategies,
- Planning capability: given a number of prepared strategies for individual subproblems, a test object might be able to combine those strategies to solve a larger problem by producing a coherent plan.
Posts
