The Man in charge of the 2nd Fastest Computer in the World
You won't believe this, but 3 months ago on my way to the SAE meeting in Detroit, I happened to sit next to the man in charge of the 2nd fastest computer in the world, Tom Zacharia. We talked the whole flight and of course, I had to ask him about Transhumanism.
The Jaguar computer, of which Zacharia is the PI, now exceeds 1 petaflop (10^15). Zacharia said that 1 exaflop (10^18) is already planned and is inevitable. He also believed that a zettaflop (10^21) will one day be constructed. This should please all you Transhumanists out there since this should be more than enough for strong AI if you believe the estimates of Moravec.
I asked him about A.I. and he pointed out that computers already do many things that people 100 years ago would consider intelligent and that we just keep redefining what 'real' intelligence is as computers advance.
I also pointed out that many people that I work with are skeptical about the usefulness of computers for modeling real world problems. He bristled at this and called those people 'cook and look' who pretend that collecting three data points and drawing a line constitutes 'understanding'. This was personally amusing because I know exactly who he was referring to. By the way, we call what the modelers do 'type and hype'.
The Jaguar computer, of which Zacharia is the PI, now exceeds 1 petaflop (10^15). Zacharia said that 1 exaflop (10^18) is already planned and is inevitable. He also believed that a zettaflop (10^21) will one day be constructed. This should please all you Transhumanists out there since this should be more than enough for strong AI if you believe the estimates of Moravec.
I asked him about A.I. and he pointed out that computers already do many things that people 100 years ago would consider intelligent and that we just keep redefining what 'real' intelligence is as computers advance.
I also pointed out that many people that I work with are skeptical about the usefulness of computers for modeling real world problems. He bristled at this and called those people 'cook and look' who pretend that collecting three data points and drawing a line constitutes 'understanding'. This was personally amusing because I know exactly who he was referring to. By the way, we call what the modelers do 'type and hype'.
Comments