AI, AGI – Is it a hype right now?
I had been following some of the AI practical use cases for sometimes. In recent years its become a big thing that every company does some AI in one form or other or that is what they tell or project and investing a lot. Lots of prominent people like Elon Musk are talking about AI's implications and its perils. Recently Googles Deepmind "AlphaGO" has been creating waves. I understand that this tool has generalized the game enough to understand the game and out perform a human which is great. But my perception is, that most of the AI that we have are oriented towards solving a specific problem and we are no were close to an AGI (Artificial General Intelligence). At-least not in the very near future. Neural networks have got very prominent in recent times as well, with these deep learning techniques and modelling some tasks around it. But these are very specific cases where the neural net is trained(with massive amounts of training data) and modeled very specifically and does its job well. But for an AGI to exist, this would be one in a billion of tasks which the AGI could perform. Also, where are we with commonsense reasoning? Without commonsense reasoning, an AGI could not make sense and interpret its reasoning at any point of time like a human. Yes there has been extensive research going on in common sense reasoning etc., but they are very primitive and has to go a long way I think. People like Ray Kurzweil have been claiming that within the mid of this century singularity will be achieved and that humans could enhance and interface with computers through human computer interface. But I think that goes in a different direction where its not about AGI but about human machine interface. Maybe that would lead to training an AGI in the future provided singularity happens and its safe to connect your brain to a computer or a network of computers.
The craziest thing is, I read in multiple books and journals that humans use only a fraction of the capacity of their brain. Hardly 10% or less. So if we humans learn how to use the other 90%, not even that, just another 10% we would have leaps on the abilities of humans as a whole which we cannot fathom as we don't know what we would be capable if we use that much more of the brain power we have and let alone going quite high up to the 80 or 90 percentile where we could be super humans with incredible capabilities. Then again what do you define as AGI, going back 10 years we would have a lot of notions non existent. So an AGI should be evolving all the time and that means common sense reasoning is at the core of it, if it ever happens. With reference to Marvin Minsky's book "Society of Mind" the thought process leads to where the interpretation of the human mind itself as a whole bunch of mind's in itself where the mind could critically analyze itself on different actions or make informed decisions before doing anything. Yes with Deepmind kind of technologies we have use of some form of AI for a specific purpose which is great, but for an AGI to exist its going to be a huge society of these Deepmind's (maybe billions or trillions) somehow interconnected and to work cohesively to a stable extent where critical thinking could happen. I am not an expert but would like to get others views and comments and interpretations on AI and AGI for the future.