Electricity outplaced steam, water and wind mills whereof the latter come back now to produce it, enabled productivity increases and education dissemination relocating industries, enabling new Technologies, etc. Similar things have started to happen with Artificial Intelligence. Nobody can stop it and all discussions about regulations and limitations will in the case of Artificial Intelligence only increase the number of human losers. Most contemporary people even don’t grasp to what extent they are already in the middle of it. Although the vision of it dates back to 1956, important parts of making it a reality today such as for example the internet were only invented less than half of AI’s age ago. But since about ten years which may be characterized as the smart-phone evolution era, Artificial Intelligence advances much faster than commonly perceived.
What intrigues me most, is actually the fact that the step from machine to deep learning had been bolstered so strongly from the gaming industry! Interestingly enough a lot of algorithms that came handy for Deep Learning had originally been developed for gaming applications. Criticizing these games to have dead beaten a lot of person to person communication most frequently between kids and parents, on the other hand brought up a generation with a totally transformed learning behavior. Unfortunately neither parents nor school education have timely adapted for this. So kids get less and less kick out of old-school learning, getting no satisfaction bonuses from it. On the other hand this discrepancy isn‘t all new. Looking at the learning of a child from being a baby until entering some sort of educational structure, we have always had this disruptive phenomenon. We humans have the ability to imaginate things from a very limited amount of perceived information. And these imaginations serve as a simulation model delivering us internal “big data“ we then can validate with reality on a “true – untrue“ or “real – unreal“ basis. Binary or Digital so to speak! Taking the word imagination from its origin “image“ the human brain is actually processing images in digital logic. Yeah, we translate perceptions into imaginations and validate the memorized depiction of our imagination against reality to form synapses for what to remember as real, good, dangerous, easy, hard, fun, boring, youname it. And this happens by putting impressions into specific sort of drawers, representing individual qualitative level planes.
Some makes us sad, others angry, fearful or excited, always communicated by our physical body‘s chemistry into all parts of it concerned or involved. Maybe certain drawers are just connected with dedicated vegetative controllers which is why synapses are being formed in places specific for the kind of learned subject matter. I would assume that these predispositions are somehow built into our genetic code and therefore inherited from generation to generation. How many synapses of our ancestors’ learnings we may also possibly inherit, is an unpopular subject contradicting the assumption that every human was equal and should have equal chances in life. But it may also be that we all start from a cleared memory, however the file-folder hierarchy, in other words the drawers, and connectivity to the various body functions being transferred from generation to generation in our genetic code? If so, we would have people with different depth and spread of memory and that might explain why some people learn certain things faster and earlier than others.
Artificial intelligence kicked really off after switching to image processing algorithms and is currently blooming with the number of processing layers connected. While Alpha Go’s break through of finding a new but valid move in a game played by humans since thousands of years, without ever coming about such possibility, was enabled by 5 layers of 50 circuits with 60 interconnection points each. In 2012 this was extended to 8 layers, then more than doubled two years later and in 2016 increased to 218 level planes. This to some extent explains the rapid emergence of AI into various new fields of applications, developing its deep learning capabilities so exponentially. Simply by trying to replicate the architecture of human brains. Therefore I do not see any reason why AI would not develop emotions in the future. To be honest, I can’t imagine how this could be prevented. Since AI learning has now started to follow the principle of childrens experience-learning, it will undoubtedly have to learn interpreting emotions of the humans interacting with it. If AI was not developed for the benefit of people it would be worthless. But to secure such benefit humans will need to work with AI and together this can bring an exiting future. So let‘s not be afraid of AI but endeavor that such machines would make us better humans. Maybe we‘ll have to educate them like our children, before the‘ll teach us what‘s up later on. Of course it would be good if by design AI could be incapacitated for destructiveness. In pervasive computing and communicating hopefully malware-manipulated units will become detectable by uninfected AI to exclude destructiveness from the networks. This would probably be the most worthwhile ethical control effort to be further pursued.