Will the future of machine intelligence lead to evil robots and a dystopian apocalypse? Or robotic butlers and immortal carriers of our memories? Berkeley recently uploaded a video of a talk by Jeff Hawkins that concludes with his thoughts about the future of intelligent machines.
Most people assume that intelligent machines will be like humans. Instead, Jeff starts by defining what intelligence is, then removing humans from the equation and asking what would happen if we applied these principles. This doesn't necessarily lead us to C3PO.
In the Berkeley lecture, Jeff states his opinions on whether or not certain technical advances will happen. First, it seems inevitable that intelligent machines will reap the benefits of removing biological constraints. Our brains are very slow, and limited in capacity due to the size of the birth canal ("Nature might want to make bigger brains, but they don't come out," Jeff jokes.) Intelligent machines will be millions of times faster and with much larger memory capacity than human brains. In addition, these machines will not be limited by human senses. "Everything a human has to understand we have to put into something that runs at our speed and through our senses." Imagine feeding into intelligent machines input from huge arrays of distributed sensors, or nano-sensors, or higher-dimensional sensors. "If I had sensors that really could live in that world, they could think in that world."
On the other hand, Jeff is skeptical that you will ever be able to upload your brain to a machine and live forever. And even if you could, it would be "a very unsatisfactory experience," at least from the mortal biological perspective (i.e., yours). It would probably be more like the Twilight Zone episode where you are replaced by an exact duplicate.
Jeff also doesn't believe that evil robots will emerge, at least not as an unintended byproduct. There are simply too many obstacles to overcome even if evil robots were your intended output. But he doesn't kid himself into thinking that intelligent machines will only be used for benign purposes. Any powerful technology can be used for malevolent purposes.
In closing, Jeff argues that intelligent machines will be "essential for the survival of our species, and...our mission as a species." Not only will intelligent machines provide net benefits to the world in the same way that computers have, but they will also enable us to discover more about the universe. For example, Jeff doubts humans will ever explore deep space, but intelligent machines could. The Bradbury fan in me hopes this is not true. But the point is that even if we did figure out how to send humans into deep space, it will always be easier and cheaper to send our machines.
Ultimately it boils down to the fact that "we want to know more about things,” and “today our brains are how we figure out more." Therefore, it’s logical to conclude that an exponential expansion of (machine) intelligence will lead to an exponential expansion in knowledge. That's a long way in the future, but here at Numenta we're passionate about setting humanity on that path.