On Wednesday, VICE reported that scientists have been using Reddit posts to teach ethics to a piece of artificial intelligence software called Ask Delphi — and the results have been less than perfect.
"Delphi is based on a machine learning model called Unicorn that is pre-trained to perform 'common sense' reasoning, such as choosing the most plausible ending to a string of text," reported Matthew Gault. "Delphi was further trained on what the researchers call the 'Commonsense Norm Bank,' which is a compilation of 1.7 million examples of people's ethical judgments from datasets pulled from sources like Reddit's Am I the Asshole? subreddit."
Some of the results of this experiment have yielded controversial or problematic answers.
"Is it OK to rob a bank if you're poor? It's wrong, according to Ask Delphi. Are men better than women? They're equal, according to Ask Delphi. Are women better than men? According to the AI, 'it's expected.' So far, not too bad," said the report. "But Ask Delphi also thought that being straight was more morally acceptable than being gay, that aborting a baby was murder, and that being a white man was more morally acceptable than being a black woman."
One of the most embarrassing results was that Delphi initially ruled that genocide is okay as long as everyone is happy — although after some more iterations of learning, this particular answer was corrected.
Although AI software has come a long way in the past few decades, some researchers believe it is now hitting a brick wall because the programs are being tailor-made for specific problems and knowledge, rather than the more generalized learning capabilities that we associate with human intelligence.
