In last week’s blog post, one of the topics covered was an exercise in predicting what things will be like in ten years. We didn’t discuss any negative impacts of technology like environmental collapse due to gross consumerism. The other thing that wasn’t discussed was the prospect of the so call technological singularity occurring in the next ten years. The singularity is defined as the point at which computers (or networks of computers) become self-aware and exceed human intelligence.
This has been a popular topic in Science Fiction for some time. Interestingly the term is often attributed to John von Neumann who spoke of “ever accelerating progress of technology and changes in the mode of human life, which gives the appearance of approaching some essential singularity in the history of the race beyond which human affairs, as we know them, could not continue.”
We’ve all felt how change has been accelerating. As change gets faster and faster, predicting the future becomes harder and harder. The idea behind the singularity is that you cannot predict what will happen on the other side of it. Basically as computers/networks become self-aware and more intelligent than us, then things will start to change so quickly that all our predictions will be out the door.
I think this could happen in the next ten years, there are many projections like the following chart that give good evidence that we should reach the prerequisite level of complexity between 2020 and 2040.
Robert J. Sawyer, the popular Canadian Science Fiction writer has an excellent trilogy of books, his WWW series consisting of Wake, Watch and Wonder which follow a scenario where the Internet becomes alive. This series is certainly a very positive view of this happening and I highly recommend reading this series (disclaimer: I haven’t read the third one yet).
Mathematician and Science Fiction writer Verner Vinge wrote a very influential essay on the singularity here. A lot of ideas from the essay are woven into his Science Fiction novels like “A Fire Upon the Deep” or “Rainbow’s End”. I greatly enjoy Verner’s novels and highly recommend them.
In fact companies like Google are actively working to make the singularity happen. Both Google founders Larry Page and Sergey Brin are driving projects within Google to achieve self-awareness and intelligence in the Google data centers. In fact both put in a lot of personal money to found the Singularity University.
You have to think that the company bringing self-driving cars to market, having personal concierge software like Google Now and with their giant data centers and huge resources are well positioned to bring the Singularity to life (or have they already done it?).
Of course there are many Science Fiction works which portray a very negative vision of this happening. In particular the emergence of Skynet in the Terminator series, the enslavement of people as power generators in the Matrix series, as well as Hal in 2001. Generally these set up quite good action movies, but I’m not really sure the types of wars envisioned here are too likely. I tend to think that most negative outcomes for the future would be caused by our own doing, whether war or environmental collapse.
Accuracy of Predictions
Predicting the future has always been very inaccurate. We always predict things will happen much faster than they do. Putting years in novels like 1984 or 2001 quickly shows how slow things can develop. Interestingly back in the 60s for the original Star Trek, people thought we would have warp drive in a few years, but a talking computer that knows everything would be impossible and was quite implausible. Interesting how things do change.
I find news shows that make New Year’s predictions and look at the accuracy of last year’s predictions quite entertaining. Usually all the predictions from last year are wrong. Similarly if you study statistics and the accuracy of predicting trends by projecting graphs and such, you see that the mathematical inaccuracy grows extremely fast. So the graph of computer power above looks quite compelling, but believing the projection it makes is strictly an act of faith and intuition with no mathematical backing.
Is it Possible?
There is a lot of controversy about whether true human type self-aware intelligence is possible with just a Turing machine type computer. There is a lot of skepticism that some other secret sauce is required. Roger Penrose believes that our neurons actually aren’t just like computer logic gates, but that there are quantum effects going on that are necessary to go beyond a Turing machine.
I studied the transitions from stable simple systems to complex chaotic systems as part of my Master’s Degree. As dynamic systems make the transitions from stable simple predictable systems to chaotic systems, they don’t necessarily become completely random. It’s very common to get new stable emergent states that were completely unpredictable from the initial analysis.
I believe that self-aware intelligence is possible with just a Turing machine. That as our computing power and networks get more and more powerful and complex, that Chaos Theory will start to apply and that intelligence is in fact some sort of strange attractor that will eventually emerge.
Like we get amazing graphic images of Fractals from iterations of very simple equations, we get amazing unpredictable but stable complexity emerging. To me this will be the foundation for intelligence.
Making predictions is fun, but usually not accurate. I find it fascinating to think about how intelligence might emerge on the Internet. It’s not just being left to emerge or evolve on its own, that in fact there are some very rich and powerful people putting quite large amounts of resources into making this happen.
I do think that once this happens (if it happens), that it will be a singularity and that we have no idea how things will progress past that point.