RE: Scientific evidence of God by an atheist (Where mankind is one likely type of God)
November 4, 2016 at 7:34 am
(This post was last modified: November 4, 2016 at 7:37 am by I_am_not_mafia.)
(November 4, 2016 at 5:49 am)ProgrammingGodJordan Wrote:(November 4, 2016 at 5:38 am)Mathilda Wrote: My eyes ... my poor eyes ....
Have trouble reading that but I did skim through it and saw something about Kurzweil. Don't listen to anything that man says.
Kurzweil is quite the inventor.
Albeit, unless one is brain damaged, technological exponential transition is observable, as his graphs entail.
He's a media scientist only out to promote himself to get funding by over selling AI and relying on a team of people to do the work for him. He's been talking about reaching super human intelligence of AI in 2040 for years now based on really a dodgy understanding of how the brain works and assuming that Moore's Law will carry on forever. I lump him in with the likes of Kevin Warwick and Sam Harris. In other words, he sells fantasies.
Sure there's been an exponential curve with Moore's Law continuing, but it's coming to an end. Exponential curves tend to saturate in reality and become sigmoid functions. The scale of the challenge is far greater by many, many orders of magnitude.
The idea of super human intelligence by 2040 relies on the idea of a neuron being a simple integration device. It's not. A single neuron is extremely complex and has more computational power than a whole artificial neural network. We're talking 100 billion neurons and about 100 trillion synapses in the brain. Each neuron has on average about 7,000 synapses. That's just for a single human brain. Each human is part of a population that has evolved over time. Intelligence requires being embodied in an environment, and that takes time. Mechanical movement is many orders of magnitude slower. Any kind of AI will require parameter searching. In the real world this happened with Darwinian evolution.
Not to mention all the other limitations such as the long emergency halting Moore's Law before it even completes, how long it takes us as humans to understand such complex systems as the brain, our ability to measure the brain, materials research and how to power such processing power. A human brain works on 100 watts. You need many megawatts to power a super computer to even get close to simulating one second of a human brain in action. Fact is we just don't yet know what form of computer will come next to take us to the next step. And the next computers will dictate the kind of solution that is available. A quantum computer that needs to be cooled to a fraction of a degree above Kelvin will lead to very different solutions compared to a DNA or biological computer.
I can assure you, I am not brain damaged. I'm experienced in the field of Artificial Intelligence.