THE PARADIGM FOR SOLVING AI IS NOT CALCULATION BUT COMPETITION
We try to solve the problem of artificial intelligence using the wrong paradigm – a problem in economics we refer to as ‘mathiness’ – when the solution from bayesian accounting to the evolution of consciousness, is not calculation but competition.
The other ‘great accident’ is the bilateral revolution, and the competition via necessity of coordination within the nervous system, followed by the division of labor and competition between the functions of our minds as predator and prey.
We don’t think of neural economy, or bayesian networks as competitions at the neural-path level, nor do we think of consciousness as the result of memory and competition in a division of labor between layers and functions. The problem of an ethical and moral machine is one that must decide a competition from the forecasting of a decision, and the negative reinforcement of decision networks that lead to involuntary transfers (harms).
I understood this from my work in the early 80’s on AI for military games. But the technology wasn’t available at the time to do anything about it. I don’t find any particular mystery to artificial intelligence other than the volume of memory and the necessity of an internal grammar or language to assist in a hierarchy continuous recursive competition between forecasts(predictions).
My open questions are the amount of memory required if we want both the benefits of mechanical memory, and the power of reason, without the economy of imprecision. I mean, look at the experiments with chimpanzee image processing vs human.
Or put differently, in order to FURTHER divide intelligence more efficiently we might very well have sacrificed memory precision for another ‘n-lateral’ revolution using language and the competition between minds using language.
The problem of continuous recursive disambiguation into serial speech requires not the preservation of state (chimp memory) for internal consideration, but the recursive passing of state (human memory) so that we can serialize state into a continuously recursively disambiguous stream of expressions between individuals. We then evolved the ability to plan from this process.
It is very hard to try to remember what it’s like to think without language. It’s like trying to measure without numbers. you can do it but only to some rather simple degree. Language is just another form of calculation. Or rather more easily understood, calculation with measurements is just a reductive form of language.
The consistency throughout the hierarchy is competition (market) between memory and perception, the competition of the neural economy, competition between neural forecasts, competition between reactions or choice of actions, competition between perceptions and minds, and ability to calculate using language (grammars).
It’s this ‘market competition’ that is the model not only for cognition, but for all of social science that results from that cognition.
We were fooled by mathiness and justificationism that results from the mathematics – the most simple of logics: the single constant relation provided by the single property of a positional name. Mathematics requires very little difference between construction and deduction. Logics break down rapidly after first or second order. Games do as well. Reason does as well. Markets(competition) doesn’t break down. -Cheers )
Source date (UTC): 2018-12-18 08:06:00 UTC