Technology is our friend: Feb 11 - If we don't watch out, Skynet will happen
Back To Normal

January 30, 2014

Feb 11 - If we don't watch out, Skynet will happen

It's no coincidence that I brand my professional activities with "technology is our friend" - because i firmly believe (especially in a business context) that digitization and new technologies will help to create better products - and to those who offer them, better business. But this does not come for free. We have to learn, adapt, create and innovate - to make technology our friend. This goes for societies and humankind in general as well, and I have the impression that we're not trying hard enough. In this case, technology will eat us alive.
We ignore that technology will replace millions of jobs worldwide in the near future - or at least we should have a much more lively debate about coding and engineering in primary and high school and an even more open debate on an unconditional basic income. And seeing that the economic system has very basic structural flaws that lead to bizarre (at best) conditions like 85 people owning as much wealth as the bottom 50% of the world population, we use technology to optimize financial markets and carve out the structural flaws even more - make them more visible and more impactful. I don't know of any serious political effort to understand the role of computing, software and programmes in creating and deepening the financial crisis. To sum it up: It will not work that way.


Not only will we run into serious economic and social trouble, we might even create our own little Skynet. Remember? Skynet is one of these science fiction visions, featured in the Terminator franchise. It was built as a defense software to avoid human error in managing the nuclear arsenal and all other military systems - but turned self-learning, self-aware and finally recognized that all humans would want to shut it down, concluding that it had to kill all humans to survive. Yes, science fiction, of course. But what we are doing nowadays with our handhelds, tablets, smartphones etc. was science fiction (at best) 35 years ago.
Look at this video (klick link):


"It takes over 2 hours and over 5 dollars in telephone fees to download the text of a newspaper".
Of course without images, and no one thought of videos these days, much less of the possibility that all of us may produce videos and photos anytime, anywhere. So 35 years from now, it is almost certain that smartphones, laptops, smart TVs, game consoles and their capabilities will look at least as ridicolous to the people as this video here looks to us.

Because today (more accurately: 18 months ago), it takes 16.000 computers to run a self-learning program that will teach itself to recognize cats - from thumbnails taken from random 10 million Youtube videos. Quote: "We never told it during training, 'This is a cat', said Dr. Dean [...], ': It basically invented the concept of a cat.' [...]" The Google brain assembled a dreamlike digital image of a cat by employing a hierarchy of memory locations to successively cull out general features after being exposed to millions of images. The scientists said, however, that it appeared they had developed a cybernetic cousin to what takes place in the brain's visual cortex."

Read carefully. "It had appeared they had developed"... reminds me of that autosuggest joke:


All kidding (memes, cats) aside, it seems that this ability of the machine was not just an accident, but way better than expected - at least they cannot really explain how the machines did that. That's my takeaway when a Google engineer, probably some kind of genious, explains that their "deep learning" systems can solve problems that their engineers CANNOT really explain or could not have created themselves. Even though the article on The Register ends with "Skynet? No", I would not be so sure of that - in the long run. Don't think 5 years ahead, think 30 years. Identifying any face from whichever angle on any photo? No problem. Identifying any object on any photo? No problem. Maybe we are close to that even today. So now add the three letters N, S and A to it. Most of what we read about them is 5 or more years old. That is yet another field where we as businesses, as individuals and, worse, as whole societies do not seem to take as much influence as we need to make the technological capabilities our friend. Machines that can learn by themselves and are already now able to extract information from a huge number of data in ways that engineers cannot explain - this means that not only the NSA seems to be out of anyone's control, but very well some day the NSA's programs might be out of their control. Sounds a lot like Skynet to me. I thought we had at least one or maybe even some decades left until that would happen - that there are machines really capable of learning (not a simulation of learning). So in the absence of a better plan than to take it to the streets and tell the governments to finally accept the unbelievable impact of digitization and technology as a whole on society, economics, business and in the end on human life, I can just recommend to support EFF.org and at least take part in the February 11 protests.