The objective of the experiment was to show how easy it is to influence any artificial intelligence if you train it on biased data. The team of researchers quickly found out that Norman wasn’t like any other normal A.I., and its exposure to a subreddit for graphic content changed the way it saw the ink-blotted images that were provided — see here. Norman’s answers just send a shiver down your spine.
(Read more)