The singularity: when machines become conscious
Tim Finin, 4:26pm 18 June 2008
The June 2008 IEEE Spectrum is a special report on The Singularity which has many short and provocative articles available online. This is what Wikipedia calls the technological singularity. The idea is that technological advances, especially those involving computers, seem to be accelerating and that at some point in the not too distant future, like in 20 years, we will be faces with the appearance of conscious machines with human or even super-human intelligence.
After all, Moore’s law says that the complexity of integrated circuits roughly doubles every two years. If we can maintain this exponential growth, and it’s held since the early 1970s, something big is bound to happen.
Of course, this is a popular scenario for science fiction books and films. As someone who has worked in the field of AI for more than 35 years, I’m skeptical. But these articles are fun to read and are very stimulating.
Ebiquity Blog
June 19th, 2008 at 4:47 pm
Machine intelligence may one day happen, but the idea that it will happen as a result of increasing machine speed is absurd. Just because you have a huge amount of processing power doesn’t mean that you can effectively mimic the architecture of the brain to use that CPU power and create consciousness.
Nobody right now has an implementable idea of real, complete consciousness. It’s just not the case that CPU power is holding us back. We just don’t know how to do it. Faster machines isn’t going to fix that.
What would fix it is somebody making a real leap to discover how the brain really works, and what the relationship is between firing neurons and “consciousness”. The brain just isn’t as simple as an automata in the computer science sense.
Honestly, does anybody believe you can simply throw hardware at this problem to fix it? You might get so much machine power in one place (one day) that you can crack 2048-bit RSA in under a second, and still not have the first idea how to create consciousness.