[jayamohan]

Interesting article, I agree with Bringsjord's position that in order for the AGI to be reliable and beneficial, one needs a sound mathematical proof to how it will operate and about its specifications. But then again, obtaining such a proof is 'probably' beyond present human abilities. Theoretically, if such a proof can be produced, one needs and some other form of AI. Now, one needs a proof to build AGI and to derive a proof one needs AGI - sounds chicken or egg type problem. But I'm sure there exists a solution.