THE ROAD SO FAR
Artificial Intelligence is a Double edged Sword - We are currently using one of the edges of the sword to tackle many of the notorious problems the modern day world is facing today full of hurdles and complicated systems but what we don’t realize is that the more we sharpen tackling end of the sword , the holding end gets sharper too until one day when we it gets too sharp we drop it and goes crazy out of hand , This feeling is often replicated by many of the most renounced names in the field Bill Gates ,Elon Musk , Stephen Hawking , and the list goes on but Science doesn’t stop for anyone , future can’t be stopped but at most only delayed but it is inevitable at a point , so why not direct in the direction where we can lead it , also a feeling replicated by all the above names which is why they are investing in leading Artificial Intelligence Companies like Vicarious and DeepMind Technologies to name a few startup's whose primary focus is improving AI and btw are doing pretty cool stuff , the aim of these companies is to develop a truly artificial intelligent system while there are many other companies and individual research organization wanting to do this they often go in the wrong direction of re-inventing the wheel when it already exist they have an alternate philosophy on AI , AI is not just computer making it play a chess game or IBM Watson defeating a human on celebrity Jeopardy , these are often misconstrued as all of AI but these are actually mere small parts of Specific AI that can only do one thing exceptionally well but when given a different task outside its area of expertise it fails miserably , the quest of these companies is to find a truly General Artificial Intelligent system like the robots portrayed in Sci-Fi Movies , they often seem to be quite a source of inspiration to inventors and developers, but to focus on creating a truly AI machine luckily we have a good news and a bad news , good news first , we already have a benchmark - a truly working AI , it is you and me , the nature's masterpiece of evolution over millions of years and all we need to do is just model , simple enough right , WRONG! here comes the bad news the brain is easily the most complicated organ in the body or any animal, more so in humans it is very difficult to model it, even the nature took 15 - 20 million years to make us even to date nobody knows the complete inner workings of brain or just put simply "Nobody knows how it works". but it is rapidly changing more and more research is done into it and attracting lot more young and bright scientist into it , making us feel it’s just down the line that we will see a scene from the movie i-robot but basic catch if you want to create an AI system you have to define it first , it should be self-learning able to figure out the problem and come up with solutions as general and different from the ones it faced earlier as possible , must be able to play chess as well as plan a mars mission , defeat celebs on jeopardy , basically be cool! so how does one get started with all this we are on the right track "Strong AI is a Combination of a variety of Weak AI together " there are too many problems in the world for us do generate weak AI for everyone and bundle and make a strong AI" so we make the strong AI self-learn all different problems , develop solutions by itself and classify and categories and stores them , repeat the process until it is truly a Universal General Strong AI although we can never be sure of this but we can be quite confident by throwing a variety of different challenges on it , moving to the next hurdle is to define "What is Consciousness?" it is very easy to get lost in the wrong direction with this answer , most people develop highly complicated sophisticated mathematical models trying emulate the process but all they attain in the end is a sequential set of instruction doing repetitive set of things over and over again , we cannot code a brain the premise itself defeats the purpose we are using a specific set of limited instructions to cover a domain of unlimited possibilities so to attack it in the right way keep reading or take a minute to think about it yourself and then check your findings
As quoted in one of Einstein’s witty answers to one of his professors question - "If God created everything , God must also have created Evil then surely he must be one of him if he created it " to which Einstein replied by giving an example "What is Darkness ? , There is no darkness in this world , Darkness merely suggests the absence of light , likewise evil does not mean presence of evilness it just means the absence of goodness , in fact evil does not exist at all and there by god did not create evil " replied Einstein while he was still in college. So forth the concept of Consciousness / Intelligence can be related on similar terms , Intelligence is not just present by itself as some magical thing which we have in our brains which makes invent all these amazing things from a simple wheel , to all the smartphones which we are such big fans of ,and all the fancy electronics as such , but given a name as we say ignorance is the absence of knowledge and hence Consciousness is the presence of knowledge and not just knowledge but the capacity of making inference out of the sea of knowledge we have in us , Categorizing and making it accessible for us whenever we need it , which helps us in making all our day to day decisions like something as simple as choosing a latte in Starbucks to something that can be as complicated as choosing a life partner , the more informed / knowledgeable the better the decisions we are most likely to make by drawing inference from the knowledge we have. Okay so now we have defined consciousness storing and categorizing is something we do often in fact the all the fancy computers are built on top it even the biggest companies Google- classifying and categorizing all the information of the world for easier access so smart programming and designing can take care of that heading for the next big ones
-The Modern day Major General Artificial Intelligence Challenges we are facing
-Developing Consciousness
-Data Storage Structure for these general self-learning systems
-Recognition , Multiple domains of sensors , Differentiation of Sensory input and feature selection , meaning there is a plethora of sensory input every second our brain gets from our 5 senses sound , vision , smell , taste and touch but our brain filters it out and focuses on more important stuff even with a specific sense there is filtering when you see you don’t pay attention to everything , do you realize the color of shirt the last person you saw was wearing or his shoes , no right, this is all information that is insignificant which brain filters and presents you with that you need to do and is pretty good at it, there is a hierarchical sense of filtering which we call feature selection, screen parsing and so on if you care about inner implementation details
As stated by Jeff Hawking’s , a leading A.I Researcher , founder of Palm Computing , founder of Numenta , a Neuroscientist , also a software Developer says - Answers to all these questions that are hidden inside us in the most mysterious and complicated organ of our body , which has more number of neurons than there are stars in the universe. Plenty of research is done and plenty needs to be done to develop a truly General Artificial Intelligent System , He also further quotes the inner workings of how our brains solves some of the most complicated problems we are finding it difficult to do in a machine , one of the keys is the work horse of brain the Cortex which models the external world around us through our senses which is fed again to Hierarchical Temporary Memory (HTM) for storage which form all our knowledge for further usage and inference if you want a coarse simplification of the cortex it looks like DeMultiplexer taking inputs from all our senses and filtering out which are the important and sending them up the line and trashing the rest. Now more interestingly knowledge which is the source of all our intelligence it is stored in HTM - this is a Hierarchical Temporal Memory where synapses , a fluid between the neurons in the brains makes connections and forming patterns and there by memory , Programmatically we need to recreate this in a machine by way of software to develop reliable memory for us to store knowledge on , the hierarchical Temporal Memory can be compared to a "Multi Level Distributed Hash Table" when the machine is just created these tables are empty just like us when are young and as and when we interact with the world through our sense we store values into these tables carving values into these table as "Experiences" which is nothing but a combination of 3 things
1. SITUATION
2. WHAT WE DID IN THAT SITUATION
3. THE RESULT OF THE SITUATION
If the result of what we did was good , we are happy , we store this in our memory but don’t think about how it all happened and go on in life but if the result was bad we carve this experience into memory and we think about what we could have done differently to improve the result and we learn a life lesson and carve the whole situation into our tables so that whenever we come across a similar situation in life again we are well prepared to do exactly what is required to come out with flying colors. this is how professionals are born they have specialized set of skills in a specific field because they have all there hash tables filled based on "Situation - Input - Result" stored for quite a large number of situations and whoever a tough situation comes along all the have to do is map this situation to the specific one in there table and they have their solution right away and they solve the problem which is also my own findings and all of this goes in to Inference part of the brain , there is no modular division of labor inside the brain as input , processing and output , you can take part of the cortical tissue in sensory hearing to sensory vision brain tissue and they switch responsibilities and work favoring hearing and vision , the brain is a system of such a huge number of sub systems but very tightly integrated among itself in hierarchy , you can imagine all of these as interrupt routines or a set of instructions but very sparsely connected to each other and each dependent on each other that is what makes it so complex everything nature design is interdependent and interwoven into each other which upon tried to debug it is impossible to read for instance the human DNA strands it’s millions of connection between just four molecules ATGC but the code is so interwoven that each connection is dependent on another connection at a different place which is what makes it so complicated and difficult to read like assembly level language of mere few simple instructions which are inter-dependent on each other which is why we use high level languages , assembly level language is a very low level computer language which is the closes we have to binary 0's and 1's the hardware understands. But the key to unravel all these mysteries is SIMPLICITY the design has to be on top upper level very simple so that when it scales it is still something implementable as Einstein said - "If you cannot explain something in few words than you probably don’t know it yourself".
The concept of developing a strong AI system is difficult because of the nature of the problem it requires knowledge from a variety of domains in a single person to realize the architecture and implementation , the vision of realizing the design of a system that can emulate our brain or at least strong AI the challenge can be daunting in itself but if you break open the inner workings of brain or anything around us ,for instance DNA our core building blocks of life is pattern of few molecules together , music - a specific set of harmonies played together , in fact anything we see around us are all made out of there similar molecules that held together by inter atomic forces , so it is all just "PATTERNS" our brain is no exclusion for this in fact all the nerve firings , synapse connections that forms are all patterns they are called Sparse Distributed Patterns , even our computers which operates on bits our core instruction set is nothing but specific patterns which our brains map specific patterns to specific actions , and no two patterns mean the same thing think of it as the first normal form for a database the nature create ,to which we have given pre-decided meaning.
We are smart because of our 5 senses which helps us understand our surroundings and our superior capability of managing and storing information and inferring this information as and when needed which is a process we happens millions of times in our brain without us knowing it which is the key to Consciousness figured out over a period of over 5 million years by our creator NATURE - the evolutionary period for first humans. There are people in this world who don’t have a single penny in their pockets and are on the streets depending on the mercy of others for their next meal and there are people whose income is more than GDP of few countries like Africa , Bangladesh , Bhutan and so on, there are people who are dumb live a mediocre life and die not making a significant contribution to science or life and then there are people like Newton, Einstein and many others whose inventions and thoughts have made this modern day possible , the question here is Why is this ? , and more importantly the answer is "Grasping Power" or Self Learning meaning Learning new things from what you already , it is a fact when we are young we have the highest grasping power we learn more and are our brains are young and powerful even when we are sleeping our brains are always looking for stimulus and rewires itself where it gets more stimulus but as age goes on we lose this power this sense of curiosity and excitement and deteriorates in everyone as opposed to few people who we call Geniuses the rate at which there sense of curiosity or grasping power deteriorates to average people is far less, Now imagine a machine where this rate is controlled by us , we can increase it when its slow and we can shut it down when it goes crazy , imagine this grasping power with the amount of hardware we can provide , machines learning at the rate of modern day processors making inferences which we can never even think of , excited a new realm of possibilities , this grasping power has already been charted down in our brains to be more when we are young and deteriorate eventually to manage the rate at which we invent and progress as a species by nature but machines don’t have to have that check
One might argue that "What about Imagination? Machines do not have feelings like us, they cannot imagine, they don’t have heart, they cannot make great music like - Beethoven, The Beetles, and many other great artists like Michael Jackson, Eric Clapton, classic legends right? , Wrong Heart is merely an organ that pumps blood , it all happens in brain , Feelings are basically set of chemical in our blood that makes brain behave in certain way forming rapid changes in cortex , neuron connections blocking the usual logical ones and favoring the different paths , imagination is a figment of reality that we perceive and our sub conscious relieves figments of these when we sleep in reality it never exists , this is a model generated during day time that is stored in our brain , and music ,in future Machines can create music and can imagine things the way we do... sounds like a sci-fi movie , not to me #IMAGINETHAT
