I read Ray Kurzweil's The Age of Spiritual Machines when it came out and although it's low on specifics, the abstract concepts are solid. The main one IMHO bei

I read Ray Kurzweil's The Age of Spiritual Machines when it came out and although it's low on specifics, the abstract concepts are solid. The main one IMHO being that once a chip can emulate the inputs and outputs of a neuron, then a human brain could slowly be replaced by silicon and the person wouldn't even be aware of when he or she became a machine. We used to say that artificial general intelligence (AGI) was 20 or 30 years away. Now it's more like 10. The main thing stopping it is the utter lack of exploration in computer science of highly parallelized/concurrent processing. So hardware-wise that means we need a DSP with perhaps a million cores with local memory (like map reduce), this would be like a video card but with general-purpose cores that can run any traditional language instead of OpenCL/CUDA etc. The software side means that we need more experience with languages like MATLAB/Octave, Elixer, Erlang, Go etc so that we can build other tools like TensorFlow on top of them from first principles. Given those two foundations, many of the methods in neural nets, genetic algorithms, etc become one-liners and kids could play around with them the way we might have written ray casters and fractal explorers when we were young. They'll quickly discover novel ways of solving problems, combining intelligent agents in various layers, automatically assigning hyperparameters, basically all the things we struggle with today. And not long after that, there will be a repo on GitHub where you can download a brute force AGI and see how fast it can do your homework. This seems completely inevitable to me and I would have loved to have made a career of it but now there's no time. Any idea you can think of is between 2 weeks and 2 years away from being manifested by someone on the web. Which is why I think we're looking at 5 or 10 advances over the next decade that will make AGI all but certain (barring a global recession like the dot-bomb followed by an embrace of fear like the global war on terror that set tech advances back 10-15 years, but I digress).