But there is a lag time between discovering a technology and when it becomes available for practical use. Especially for things that involve the human body, such as drugs and implants, where the lag time is easily 10 years just to go through all kinds of trials and approvals.
I think this time frame is too optimistic, since what you're talking about would amount to quite revolutionary changes in less than ten years. A general rule of thumb is that science/engineering projects always take longer than you imagine, since a human mind isn't capable of foreseeing all the possible details and problems. This doesn't mean that I don't think there will be exciting developments occurring within the next decade - there will be plenty - but on this time scale I'm not expecting to see superintelligence created (at best some early precursors may appear, or the range of AGI designs may be narrowed down to a few successful ones).
A lot of singularity technology doesn't have to be available for practical use.
For instance a computer capable of simulating a human level intelligence would only need to be made once, regardless of how much cost or effort would be required, once you have one you can just double the number or processors and get super-human intelligence or wait 2 more years and replace the current ones. Alternatively if you have enough storage space you can probably run a super-human intelligent at half speed on the hardware of the time. If the super human intelligence works (ie doesn't try and kill us all, isn't insane, can communicate, etc...) if can tell us how to develop better technologies, and get them implemented quicker.
Another example would be a molecular assembler, once you have one capable of reproduction it would be hard to stop it getting out, all you need is one person to sneak it out. It would be very hard to stop someone sneaking out something a few atoms big. Once you have a molecular assembler you have something capable or making as many processors as required for AGI, and they would likely be more efficient since you can build them from atoms.
As for the basic medical stuff, if its blocked officially it will be driven overseas or underground, this already seems to be happening with bigpharma patenting medication and charging excessive amounts for live saving medication. See how long the rich are willing to wait for immortality drugs, or how long another country would take to realise that they could start to create geniuses that can design better weapons, DARPA are already funding BCI's for 2009.
Non regulated, regular technology seems to develop as fast as a way to mass produce it and a market will buy it. Although patents and legislation can slow down development a lot. Thinks like e-Ink displays, new processors etc... don't need to pass a large number of tests to be implemented, eink was only anounced a year or so ago and you can already get devices that use it.
> a computer capable of simulating a human level intelligence > would only need to be made once
Right, but getting there is going to take time. That's what I'm referring to as "practical use". I'm not talking about time 'after' building a computer with human-level intelligence, or time after building the first molecular assembler. I'm talking about the time it takes to reach there, which is unlikely to be in the 2012-2018 range that the author suggested.
Of course there is also the issue of definition. For example, when will we have BCI? The answer is that we already do. Cochlear implants and deep brain stimulation, among others. So, it's tricky to define a time-line unless we precisely describe what we're talking about.
brianwang2 votesThursday, May 22, 2008 at 11:57 AM
As the author, I will define what I mean by singularity significant. I mean that there will be significant economic and social impact from what are currently termed singularity technologies. [AGI, BCI and nanomedicine]
One generally human level intelligent computer would mean that machines with one tenth or one hundreds the power focused on economically significant tasks will be fairly common.
With new Cell processors, FGPAs and GPGPUs we have teraflop and multi-teraflop class workstations and computers common for researchers and engineers now. Petaflop is top of class and uncommon.
In the 2012-2018 timeframe, I believe there will be exaflop class top level machines and petaflops common. 100 billion neurons equivalent for specialized top level machines. one hundred million to ten billion neuron systems relatively common.
Many jobs do not require much improvement in current systems for automation/early AGI to displace the people in large numbers. The difference in staffing levels of Web 2.0 companies versus Web 1.0 companies. Automated call centers versus banks of operators. We are seeing this now with leading edge people use VOIP and other technology.
Robotic vacuuming now but more machines for dedicated mundane tasks.
My articles about biomarkers (which require nanoscale analysis of blood) and protein related imaging (two stage cancer detection), I consider next step nanomedicine. Effective, cheap and widespread detection can like pap smears for cervical cancer cut cancer deaths by 70-90% for some types of cancer. Saving tens of billions in the US, hundreds of billions worldwide each year and lots of lives. http://nextbigfuture.com/2008/05/biomarkers-for-bloodtests-and-protein.html
Nanomedicine is helping to enable effective gene therapy. This will first be deployed overseas. China and other countries already are allowing gene therapy and stem cell treatments. More medical tourism.
In the 2012-2018 timeframe, I see effective deployment of myostatin inhibition. A safer than steroids way to boost physical strength and improve health. Tens of millions of people.
I see Sirtuin/calorie restriction mimicing deployed.
Several big impacts will be rolled out such that the case against AGI, BCI and nanomedicine will be demolished. The discussion will be when not if.
On the BCI front, more direct connected to the brain prosthetics, robots, games with higher bandwidth and more sensitivity and capabilities.
Also, if DNA nanotech and some other technology can be effectively bootstrapped. We could be far closer to the full real deal Singularity.
Wal is correct that it would be another 10 years before everyone owns or has access to their human level AGI.
But we would be into the 1996 Internet equivalent phase by 2012-2018.
