To think is a very complex process no matter how small or idiotic or great the thought is. Acting on the thought is further complex because it becomes a selection of choices. Artificial Intelligent programs are supposed to do both. Thinking in AI world means processing the information while acting becomes executing subsequent program or to produce a real world action.
Baby Vs The RobotIf you have spent your hours and days trying to build an AI bot you would have by now definitely appreciated human brain for its beauty of functioning. Human brain is one of the best creations of the nature. The comparison between AI processing and human brain comes when the real world problems cant be solved with just mathematical equations and rather needs a logical perspective. There is no mathematical equation for LOVE!! or maybe there is 2×idiot=love ;). Keeping the joke apart even the brain of a newly born knows who the mother is whereas AI codes needs tremendous amount of energy and logic for facial recognotion. The story does not end here the brain knowing the mother acts accordingly when hungry or want something but again for AI code it comes down to more learning. The point here is if we go by all the things that even an average person gets the training for during his/her life then its very time consuming and difficult to train AI program for all those things. A simple example would be as below, see the two versions of writing a word,
Version 1 : |_O\/£
Version 2 : LOVE
While in an instant you would have read both as "love" the AI code would not have done the same for the 1st version unless specifically designed for it because almost all letters are special characters. Although you had a better context and you knew whatever you are reading is a word but the AI code might not have that.
What is the actual difference?Lets look at some facts about human brain and the computational powers that AI codes rely upon.
1. Memory or Memory 1,0,1Human brain: Genetically programmed to create, store and process relations between two entities based on associative memory/context
AI: Maintains and processes relations based on occurances/provided data.
Example -Consider a situation where a person greets you stating he/she knows you from blabla.
Now you try to remember but cant or maybe after some time you do. But have you ever wondered why the other person remembers you but you cant? Apart from lets say having a weak memory its because you did not associated the context or memory to that person back then. But for AI once a record gets in its store in this case the person does exist otherwise not. There is no grey area in AI even if its inserted programmatically it cant be processed as an actual information so ambiguity is a no no for AI.
2. Understanding vs FactsHuman brain: Learning is based on assumptions, facts, experience
AI: Learning is based on data
Example-Consider that an intelligent program and a person is given a book of Harry Potter to read. So unless you explicitly mention that Magic is not real the program will process everything as real but an average person would based on experience and assumptions only will mark it as a work of fiction.
The main thing here is in real world there is many false/complex data which may cause an unintended machine learning and its too tedious a task to get the correct data with proper context and leaving no choice but to rely on a judgment call which might cause a disaster for coding.
3. Machine Learning Or a School?Human brain: Does not requires millions of data to understand something
AI: Requires similar data to understand a pattern
Example-Remember when way back in your school you haveto learn ABCD..Z? Well your teacher did not had to draw thousands of A's in different shapes ad formats to make you understand the same. They just had to tell you what A is and how it basically lools like. After knowing that you can understand the letter A even if it was written in different shapes/fonts in a word. E.g.
Artificial Intelligence has to undergo machine learning process and NLP to grasp this. In fact they are few of the major challenges while building an AI systems.
There are so many differences between a human brain and an AI processing that they cant be covered in a single post. But for starters these are some ofthe major differences. I am waiting for a day where an AI system can capture the images from a camera in goggles, converts them to electric impulses and transmit to neurons so that they can act as an eye for a blind person. This will be the day I will actually start fearing AI. You think AI can achieve this? Don't forget to comment, subscribe and follow on official thunkmine.com channels and add to the thought process.