Artificial Intelligence - The AI Revolution

“By far the greatest danger of Artificial Intelligence is that people conclude too early that they understand it.” 
― Eliezer Yudkowsky

A brilliant TED Talk by Nick Bostrom

For more details on AI - Read the brilliant post here

  • Three types of AI in the works - Artificial Narrow Intelligence (ANI), Artificial General Intelligence (AGI); and Artificial Superintelligence (ASI)

  • The misconception about AI is that they are robots - not true

  • Robots are only external containers, AI is the computer chip inside

  • ANI- works today i.e. Siri, Integrated home solutions, Traffic signals, IBM Watson, Self Driving cars are examples

  • AGI - is where the computational power equals that of the brain. Currently this is difficult to do because it costs 390 million USD, and is currently exist only in China (Tianhe-2) Google is investing billions in optimizing AGI

  • The other challenge is replicating human intelligence - three ways to go about it - 1. Plagiarize the Brain 2.Teach it and feed it which humans naturally evolved over the years, 3. Simulate the environment for it to learn on its own, It reaches its critical point and picks up on its own

  • When there is an inflection point where the AGI turns into ASI and that is the scary part, which it will surpass human intelligence at an accelerating rate

  • The law of accelerated return - the amount of technological growth that we are seeing is exponential i.e. what took 100 years in the past is being done in a decade and in the future it will be headed to being done in a month - eventually reaching a point of singularity