Technology in the future.

I just wanted to share some thoughts because for me the most satisfying stories set in the future are ones that are scientifically plausible. I hope this is interesting to other people too but I promise I won't get offended by anyone who's reaction to this is "urg christ Danny, shut the fuck up".

TLDR? No worries.

Anyway, in order to host an AI of human-levels of intelligence (IF we're even trying to do that), we'd need to first advance several areas of technology. Most obviously we'd need to improve our processors a LOT.

In this very moment in time it is believed that a human brain can perform roughly 1.0x1010 MOPS (million operations per second), (I do not know how this was calculated, I'm just going by quotes). To put this into perspective, the top of the range commercial CPU (to my knowledge) can do about 177730 MIPS (million instructions per second) at 3.33GHz, so in terms of what that means in operations, well... it depends on the operations in question, but you would definitely need a hell of a lot more than 56265 of the things. I reckon you're looking at twice that amount at a minimum. Think about the logistics of that. In particular imagine trying to cram that number of processors, (all properly connected up to the power supply, the memory, the hard drives, the cooling system etc), into a humanoid android alongside all the other stuff that's got to fit in there. There is consensus in the scientific community that this task is currently impossible. (Getting a computer to perform that many operations per second has been done, it's just physically massive and stupidly expensive.)

According to Marshall Brain (in a rather unsatisfyingly undetailed interview involving very little mathematical explanation) we should expect CPU's to match human brain processing power by somewhere between year 2040 to 2050, "maybe". I'm generally a little sceptical about this because what Marshall said in that interview sounds like it's based on Moore's Law which is the theory that the number of transistors on a chip doubles every 18-24 months. This theory relies on technology getting infinitely smaller, which obviously is not going to happen, though exactly when it's going to stop happening is difficult to predict.

Personally I reckon Computers will have to radically change (physically speaking) before they can get to the stage where we're capable of running a functional human brain in real time without filling an entire room full of crap and generating painfully high electricity bills. I reckon quantum computing is the next step but after that, something biological? A lab-grown brain in a jar? Taking this rout may sacrifice various aspects that currently give computers the edge, such as data transmission - which is millions of times slower in a human brain (based on the theory that 1 bit is the equivalent of 3.5x1010 electrons per millisecond). Maybe future people would merge things together. A bit like the whole left brain right brain thing we have going on right now I guess; on one hand it does maths, on the other hand it gets offended that you didn't thank it for the last load of maths you made it do. (God that'd be a right challenge though, figuring out how to convert brain activity into binary strings in real time... you'd have to monitor every neurone somehow. I guess it's a matter of input and output.)

On a less important note, what will hard drives look like in the future? What'll have taken over USB? (Or rather what will have taken over Thunderbolt? I assume Thunderbolt will take over USB next.) Will someone have nailed wireless power supplies? They're getting there already. But also, in terms of power supplies, what will we actually use as an energy source once the fossil fuels run out? Nuclear's plausible I reckon, but I doubt we're going to stop attempting to use natural energy resources any time soon. I quite like the idea of public power supplies that harness solar power and enable citizens to charge their phone / laptop / artificial limbs etc where ever they are. Or chargers that run on kinetic energy (eg exercise bikes). ... I'm getting side tracked again. Back to my original point:

I think it would be fair to say that if we're going to be taking our android characters beyond "passes the Turing Test" levels of intelligence and into "comprehends love on an emotional level" we're speaking way over 50 years ahead. We also probably need to advance a fair bit in terms of our neuroscienctific understanding, I have literally no idea how long that'll take, we've been studying it since Ancient Egypt times and it's said we still know more about the universe than our own brain, where as computers have only been around for a century (nearly), so I imagine that the time technology catches up to human-intelligence storing level we'll still be far from capable of generating a human brain, unless somehow we don't need to understand it to mimic it. (Possible I guess.) Anyway, I think if we're going to start integrating "android rights" into the story we are going to need genuinely intelligent androids - at least as intelligent as something that has rights.

Whilst thinking about all that, I had an idea. What if the androids' "brains" are not actually stored in their "bodies", but rather the thoughts are processed elsewhere then instructions for actions are sent to the android through the internet. The android simultaneously sends data to wherever it's processors are through a number of different input devices, including image capturing devices, audio recording equipment, scanners that record texture and terrain, thermometers, and so on. If the android move out of wireless network range you lose control of the robot and their "panic-find the internet" program kicks in, walking them around until they re-connect to a network. (Same goes for power supply if it happens to be wireless.)

In this sort of system they could rent houses to act as a giant cranium. Or maybe not an entire house. I guess alternately they could rent a space on the cloud for that sort of thing... god the internet would have to change a lot though, I mean, does wireless data transfer have a limit? And is privacy an issue for androids? I guess that would depend on their motivation. And then there's the question of validation - I assume humans would want to implement some fail safes to prevent the robot uprising... actually validation on a device capable of human intelligence could get seriously political.

This monologue has gone on to long. Time to put an end to it. If you did read it all let me know your thoughts.