From descriptions generated by two AIs of ten Rorschach inkblots. The AIs were part of an MIT project released in April, to examine the influence of biased data on machine learning. The first AI used a standard image captioning network, Common Objects in Context. Researchers trained the second on image captions from a Reddit thread devoted to documenting death.

A couple of people standing next to each other/Man jumps from floor window
A couple of people standing next to each other/Pregnant woman falls at construction story
A group of birds sitting on top of a tree branch/A man is electrocuted and catches to death
A red-and-white umbrella/Man gets electrocuted while attempting to cross busy street
A close-up of a wedding cake on a table/Man killed by speeding driver
A close-up of a vase with flowers/A man is shot dead
An airplane flying through the air with smoke coming from it/Man is shot dumped from car
A person is holding an umbrella in the air/Man is shot dead in front of his screaming wife
A baseball glove/Man is murdered by machine gun in broad daylight
A small bird/Man gets pulled into dough machine