A review of DeepMind efforts in Artificial Intelligence

#1  playing GO   see minute 24

#2   imagination in computers

Published on Apr 27, 2017

In Nov 2015, Google's DeepMind's founder Demis Hassabis lectured about the future and capabilities of artificial intelligence.

31:30   Rat dreaming ( imagining the future )

37:00   Computer dreaming    here the computer imagines a future game

#3   Google using AI to provide services
           analyzing photo images


These videos suggest to me that the real challenge is to choose the best topics to focus AI Deep Learning on.  I would suggest one topic would be the so called justice system and in particular *Prison* ...... Deep Learning requires Big Data.   So the issue would be .... what data does DL need and where and how will it be gathered ?  This is a deeply challenging philosophical issue.


The current prison system in the United States ( and the systems that surround it ) are a gigantic failure.  The systems do not work.  They seem to be getting worse in some cases.  The typical supporter of these systems insists "We have to have something".  "Something" in this case is a failure and supporters insist that this failure is better than having nothing.

AI can suggest a better approach for the future.  But it will be incredibly difficult to define to the AI what "successes" look like.  It will be incredibly to collect "useful" Big Data.  Bias from all the participants in this system will be massive.  Playing the game of GO is simple by comparison.


 At in 50:00 video #2 someone asks if AI will screw humans over ?   Here is my question ...... "Who do you fear more and expect more abuse from ? ...... Big brain scientists and entrepreneurs .... or less intelligent politicians, bureaucrats, regulators who have command and control agendas ? "  

Hitler had an estimated IQ of 125.  According to Harvard professor Henry Murray , Hitler managed his insecurities by worshiping "brute strength, physical force, ruthless domination, and military conquest."  Hitler apparently felt the German people must conquer the world ....... or be eliminated for their weakness.

Would an ASI have these or similar motivations ?  I doubt it.  I think the greater the intelligence and the greater distance from an Amygdala-like processor , the greater the chance that an ASI would be friendly, fair and easy-going.



tags:   99guspuppet    777DeepMind   777DeepLearning