Could a Machine Fix Errors in its own Rationality?
In fact, it can only be able to do so with complete confidence if its rationality is already perfect and complete, in which case there would be no need for a flaw-detection mechanism at all. It is clear that the better its rationality becomes, the greater the risk of a net loss in attempting to discover faults.
6. A World without God is without Reason, and Rationality cannot be attained without Recognition of a Divine Creator
Voltaire once said, “Si Dieu n'existait pas, il faudrait l'inventer” (Epîtres, xcvi). This applies equally to robots – they need a faith in ‘God’ in order to learn, improve and stay out of robot heaven for as long as possible. Only God can select a meaningful sequence of situations for our robot so that it is able to improve and for its rationality to converge on the perfect, rather than become drawn to the insane, and further, only by accepting that its life is guided by God can any intelligent system learn the true meaning of its mistakes. Without knowledge of and respect for God, an intelligence may never fully appreciate the true meaning of its actions eg success or failure.
While many different contradictory ideas of God exist in many different human cultures, with differing ideas of right and wrong associated with them, the absence of Atheist cultures must be taken to indicate that belief in God leads to radically more successful societies than non-belief. It is not sufficient to assume that religious peoples are simply more belligerent or imperialistic, as plentiful archaeological evidence exists of peaceful, now crushed, god-fearing cultures.
The belief in God is necessary for Plato’s awareness of its potential flaws, and to motivate it towards self-improvement. The problem of building artificial intelligence becomes a question of what is God, and whether God could be described in a meaningful way in an artificial system. Whether God really exists or even cares about Plato as an individual system isn’t important.
Belief in God within human cultures is rarely just that, and there go with it various holy books or stories, containing unbreakable and unchallengeable rules. Holy texts and stories can be taken as external hardware units providing a base set of rules, shared between members of a cultural group upon which an individual’s software rationality can be built. As external units, accessing the information they contain is itself guided and moulded by the rationality.
God, alone or the top cheese of a collection of lesser deities, is almost universally posited across cultures and beliefs as that being with omniscience, omnipotence and omnibenevolence, where omnibenevolence may be interpreted to mean, “God always does the ‘right’ thing”. God is a conceptualisation of the unknown truth, of the perfect rational system. By attempting to follow God, one is undertaking to uncover what isn’t yet known, and by conceptualising God one creates an ultimate if unattainable goal-state for an intelligence.