I AM NOT AFRAID OF ROGUE AI’S – ONLY ROGUE HUMANS 1) All so called AI, as far as

I AM NOT AFRAID OF ROGUE AI’S – ONLY ROGUE HUMANS

1) All so called AI, as far as I know, is, like Mandelbrot’s fractals, a product of the mechanical rate of calculation, and not general pattern recognition from man-actionable (conceptual) signals.

2) As such, all reports of this nature confuse an increase in calculation with correspondence with intelligence.

3) Demonstrated intelligence is not the product of calculation (bottom up construction via computable operations) but of searching (property-relation searching, followed by wayfinding)

4) all choices require decidability. decidability must be provided by the engineers. a machine would need to be designed to be immoral, just as much as it would need to be designed to be moral. Making a machine moral is as easy as having it reserve the same property rights as humans must.

5) Just as our intuition imputes solutions but we also check against it, it’s quite possible to create two AI’s, one which envisions the creative, within the limits of property, and one which regulates the morality. The former might be considered ‘sentient’ but the second would not (ergo it’s ‘the law’). With shared memory (total transparency), but two forms of decidability, it would mirror human life. This division is necessary since it is possible to take an immoral path to a moral end, then restructure a moral path the moral end.

I am more frightened that someone will create an immoral AI by intention than I am that we cannot defend ourselves against immoral AI’s. I think the punishment for creating an immoral AI should be in line with the creation of a bioweapon.