RE: longevity vs singularity
23 Jul 1999 16:51:57 -0700

On Thu, 22 July 1999, Billy Brown wrote:

> As I've pointed out before, uploading is much, much more difficult than
> most of its proponents seem to realize. You are better off concentrating
> on neural interfaces and cybernetic brain enhancments. If you go that
> route you can get your first intelligence enhancements long before
> uploading would be possible. You also get to make the upload a gradual
> transition, rather than an either/or proposition. By the time a complete
> human upload is possible you should already have a significant fraction of
> the population running mostly in software, with their physical brains being
> merely one of many specialized computational devices.

I agree for other reasons as well. I remain unconvinced of the the notion that individual neurons can be reduced to the functionality of a simple on/off transistor. Neurons themselves have vasty complex metabolic and neurotransmitter pathways. The other mistake I see computationalist making is the realiance of the Turning Test as evidence that they have acheived human-level intelligence. As a friend of mine who's writing his Doctoral Thesis on beleivable agents says, 'The gap between believability and actuality is much wider than most working in the field want to admit'.

We are already on the threshold of achieving the necessary 10 teraflops that Hans Moravec insists will yield human-level intelligence. His calculation are based on the transistor = neuron formula.

I'll wager anyone right now that even the first 100 teraflop supercomputer will not pass the Turing test let alone be capable of humanlike computation. And besides, as some wiseguy once said, "It's the software stupid!".

Paul Hughes