A Turing Machine was something I never got to in Comp. Sci.
They did mention Neural Networks!!! XD And those are cool.
I suppose the Turing Model is a way of refining the Neural Network, besides using the method I understand (Genetic Algorithms).
And apparently, she's found an improvement on the traditional Turing Model.
The only thing I don't understand about it, is how they define success?
When engineering a neural network, whether by Genetic Algorithms or other mathematical processes, you have to have a way of measuring success.
This tends to create a network that achieves a specific purpose very well. Their article made the AI they're developing sound very open-ended, but I have a hunch it wont be. Or if it were, at best, it would exhibit the ability to be trained (which would be no small feat). Though, I question how it would be trained. With animals of this world, their instincts give us a basis upon which to mold trained behavior.
A creature without instincts, even if you start dolling out rewards for behavior, I doubt it's going to work without some type of foundation to begin from.
I guess it could work if you programmed some innate instincts, which are essentially, innate measures of success. Breast feeding from a mother is something a baby exhibits. If a similar instinct could be fashioned into a trainable network, you might have something.
Let's make bets of how many years before we end up with Robot babies that we train into adult robots!