Code cannot be claimed to be secure unless it has been designed with secure design patterns - patterns for which there is an "assurance argument". If the code was "coded" instead of designed, then there is no hope of creating assurance arguments after the fact. In that case, the audit will be very difficult and untrustworthy.

The question is confused. The AI processes that are thought to be most promising with respect to eventually leading to intelligent machines are not algorithmic. That is, they are based on neural networks. And while it is possible to simulate those using algorithms (e.g., the RBM cascade pattern, with each node simulated as an RBM node), these "algorithms" are non-deterministic - they are simulations - we cannot know the outcomes, just as one cannot know the outcome of a human's thought process. But with regard to religion, that seems to be a human predisposition with a genetic basis, and since religion is based on absolute unquestioning faith in things that are seemingly preposterous - a cognitive dissonance if there ever was one - I don't think that we can expect intelligent machines - which are free of our cognitive impairments - to be susceptible to conversion to religion. Still, I suspect that artificial intelligences will have the same wonder and confusion about existence as we do - they just will not look to iron age shaman texts for their answers.

Above all, take a profession that is what you love to do - no matter what income it produces (as long as you can get by). And stay focused - don't let time fritter away. Make a plan for your dream - a practical plan, for which you know all the steps, because all steps are simple and very doable - and stick to that plan no matter what.

The assumption that "we will probably find new things to do to satisfy limitless human demand" has no substantiation. In fact, the rate of job replacement is now higher than the rate of job creation. Unemployment will be rising inexorably and at an increasing rate.

Perhaps it could be like the real thing, if done right - but the real thing throws you around, and you can fall off and roll onto the ground. To duplicate that would be quite a simulation! But as you say, if it is raining out, it sounds like the next best thing.

What is driving the rush into these things is that the companies behind them want to mine the data that they generate. Imagine: all those devices phoning home, and companies able to collect data on when people are using things, where people are etc. - all that data can be input to data analysis and find patterns. Those patterns are worth a-lot of money.

It is nothing like real. Measure the calories expended - it will not be comparable at all. This reminds me of one time when Konan Obrien challenged Serena Williams to virtual tennis on the Wi, and Konan won. Yeah that's real tennis - not. :-)

LOL. Well you have discovered that most developers know very little about application security. And here we are, wondering why things are so insecure - and heading head-long into the "Internet Of Things". What a train wreck that will be unless things change. Read my article about this: http://www.transition2agile.co...

You are right that we need machine learning. It is possible that the class of learning algorithms invented in 2006, known as "deep learning", can do the job. See this TED talk: http://www.ted.com/talks/jerem...

I certainly agree about including physical education, etc.

Perhaps we are the same generation. I am 58. I too have heard for a long time that we would have AI by now. But just because it has taken longer than expected does not mean that it is not coming. Curing cancer has taken far longer than we thought, but I hope we will eventually figure it out - perhaps in the next decade or two, since we now are able to virtual experiments by simulation and manipulate genes a million times more rapidly than we could a decade ago. So I think a cure for cancer is coming, and I think that AI is coming.

But I don't think that human-like AI is needed to write programs: I think that "deep learning" algorithms can do the job: they just need to be trained, and they need a human to guide it. I think that we will see programming teams disappear, replaced by a "learning system operator" of some type. That is probably a decade away, but my assessment of the potential of this class of algorithms, invented in 2006, is that they can write programs. Perhaps I am wrong. If I am right, then we already have the technology - it is just a matter of refining it and training those systems to write code based on descriptions of a problem - the same way that IBM's deep learning system learned to play jeopardy.

Your point is right however, that if we can replace programmers, we can replace a great many jobs. That is in fact the chief concern with these systems. Please see this TED talk: http://www.ted.com/talks/jerem...

;-) The compileatron that created me was my K12 and subsequent education. I suppose the programmerators were my mother and father.

LOL. Clever. I think the resolution of that is that there is no guarantee that the deep learning system (or indeed, a human) will be able to write the program requested by the users... ;-)

It won't be long before deep learning systems are taught to code. Coding is a dead end. Teach kids fundamentals - math, science, writing.

"whatever we do, I want the kids to obtain marketable skills."

Really? By the time those kids are out of college, programming will be obsolete.

My main requirement for programs today is that they don't make me wait too much.