You might think that movies like the various sequels of The Terminator are completely science fiction with no relation to reality. But if you read about the concept of the Singularity, then you would start believing that it's indeed possible to have super-human intelligence in the future that would have the ability to create yet more intelligent machines.It would be difficult to predict the future beyond that.
Is it really possible? We had a chance to interview the man who coined the term “Singualrity”-Vernor Vinge, who had predicted in 1993 that the Singularity would happen in the next 30 years. We interviewed him to find out the causes and indicators of the phenomenon, the role of free software in getting there, and much more.
Q>You first spoke of the Singularity 18 years ago. In your 1993 paper (http://ld2.in/3k5), you said we'd get to the Singularity in the next 30 years. 18 years have passed. Are we on track?
Vernor: Things still seem to be going on that schedule. In the body of the paper, I say I'd be surprised if the Singularity happens before 2005 or still hasn't happened by 2030. I still feel that way, but with the proviso that catastrophes like a general nuclear war could easily prevent or postpone the Singularity.
In case I end up being surprised, I have another essay (which also discusses some software issues): "What if the Singularity does NOT happen"
Q>What are some of the indicators for the Singularity? How do we know when we're there?
Vernor: Having indicators to track is one of the most important reasons to think about different scenarios. In case of the Singularity, I see 4 or 5 different trajectories that could end up giving us superhuman intelligence (see "Signs of the Singularity"
First indicator is sustainment of Moore's Law, but just as important is hardware progress that is only indirectly related to component density, but which may more than sustain the computational trends even as component densities hit atomic limits.
One negative indicator is software failures, especially with large projects that fail to manage complexity.
Another indicator is for the AI (Artificial Intelligence) trajectory. Two milestone events can be considered here. One is the second DARPA (the US Defense Advanced Research Projects Agency) automobile challenge, which invites leading brains in the US to develop unmanned vehicles. Second is IBM's Jeopardy playing supercomputer “Watson”. One way of knowing that they are milestones is if critics feel compelled to respond with, "But that's not really what intelligence is all about."
Some other indicators include neuro-scientists observing biological minds discover winning algorithms that can be applied to computer systems; Computer systems that produce real inventions; Prosthesis for parts of the brain, even just a few hundred neurons.
Another indicator is for the Digital Gaia trajectory, which looks at the rate of spread of embedded networked processors into the things we make. The spread of untethered networked processors into the environment.
While there are lots of other indicators, I'll sum up with indicators for the Internet empowerment of humanity trajectory. Does the present wave of Internet mediated social upheaval lead to more peace and freedom for people, or do the upheavals turn into replays of George Orwell's novel “Animal Farm”? Another indicator in this trajectory is the diffusion not just of technology, but of the _source_ of significant inventions, so that the people are producing innovations of the magnitude of great research organizations back in the twentieth century.
Q> There are large communities of developers contributing open source and free software for the rest of the world to use and improve upon. Is that playing a role in bringing about the Singularity?
Vernor: Yes, in at least two ways. One of the trajectories to the Singularity is the Internet empowerment of humanity. Along with intelligence amplification of individuals, it is a trajectory in which we humans can "ride the exponential" and have some chance of increasing the safety of the process and influencing the nature of the other trajectories toward superhuman intelligence.
The free software movement is very important in this connection. In our billions, the human race has thousands of different _types_ of genius. In collaboration, that variety of genius trumps all the talent and expertise in governments and corporations. The free software people can provide the p2p, bottom-up technology to sustain that collaboration.
The most recent event like the Singularity was the rise of humankind among the animals. But I think the best analogy to the Singularity is the Cambrian Explosion of half a billion years ago, when complex lifeforms arose in just a few million years. By that analogy, we are now compressing the equivalent of millions of years of biological evolution into just a few decades. In the run-up to the Singularity, the physical nature of reality will likely become as volatile as electronic financial markets are now. In that environment, global centralized solutions are really just a guarantee of global profound disasters. On the other hand, heterogeneous solutions, especially where devices can still function without support from clouds, give us a chance at fire-walling ugly surprises. This is a very important value of the free software movement.
Q> Any examples of free and open source software accelerating the pace?
Vernor: I can think of items that aren't necessarily about accelerating the pace, but about free software's role in making the pace safer and more sustainable. One is computational power trends in consumer-available devices. If the lag between the supers and the home devices increases, that could indicate that independent work on the most powerful apps will also be delayed. Second is the appearance (or not) of free software for voice recognition, OCR, natural language translation, whole body UIs, real-time collaboration and advisement...That is, progress toward having the building blocks for really cool empowerment.