Vor ein­er Woche schrieb Mike Mac­Girvin, Erfind­er von Frien­di­ca und Hubzilla:

The big prob­lem is in the train­ing data and “goals” of the deep learn­ing process­es. The algo­rithms have been cre­at­ed by humans work­ing for cor­po­ra­tions with a goal to dom­i­nate the mar­ket, max­imise prof­its and share-hold­er returns, to pry into the per­son­al moti­va­tions of the observers in order to manip­u­late them more effec­tive­ly, and vio­late all known laws and stan­dards of eth­i­cal behav­iour to do this. The algo­rithms are judged by their abil­i­ty to do these things and infe­ri­or algo­rithms dis­card­ed or reprogrammed.

Heute bestätigt Twit­ter:

Our ulti­mate goal is to encour­age more free and open con­ver­sa­tion. To do that we need to sig­nif­i­cant­ly reduce the abil­i­ty to game and skew our sys­tems. Look­ing at behav­ior, not con­tent, is the best way to do that.