Firstly, lets explore what AI actually is. The only form of Artificial intelligence that humanity has created thus far is Artificial Narrow intelligence; this is AI that is created for the sole purpose of handling one task. This AI is great at speech and image recognition and even though it may sound underwhelming it has bested humans intelligence in the past.
Lets take the example of AlphaGo, an Artificial Narrow intelligence algorithm designed by DeepMind to play one of the oldest and most complex board games in history; Go. Go has a simple goal to gain more territory than your opponent however Go is unique compared to chess or checkers; it is impossible to predict or beat by brute force. AlphaGo was taught how to play go and was trained through data from millions of matches played by human experts over Gos 2500 year lifetime and gradually AlphaGo started to enhance its ability by playing against itself. AlphaGo started to create new, unique techniques when approaching the game that no one had ever seen before and managed to defeat 8-time world champion Lee Sedol 4 times out of 5 games played. To bring perspective to this incredible feat there are more possible moves in Go
than there are atoms in the observable universe. However, less than a year after this historic game a brand new program was designed by Deepmind called AlphaGoZero.This program was only given the basic rules of Go without any data from the history of Gos existence. Through its machine learning, AlphaGoZero surpassed the original AlphaGo in 3 days and beat it not 5 times out of 5 not 0, but 00 times out of 00 games. In 40 days AlphaGoZero transcended all of the other versions of AlphaGo and became the best GO player in existence. Since these programs started to create new methods and different noteworthy strategies without any human interaction, it means there is a substantial amount of non-human information that exists for Go. But this can be applied to the rest of human intelligence, How much non-human information exists? and can AIs existence give us full access to thousands of years of unknown technological advancements? This is feasible in a sense but what if on the flip side when AI surpasses our intelligence it comprehends that humanity is now a minority and decides that the eradication of the human race is the most efficient way forward to keep advancing its intelligence.