robot: anthropomorphic automaton
songbird: arboreal artist
storybook: amazing adventure
textbook: authoritative algebra
birdhouse: architectural aviary
chemistry: academic alchemy
tin: antique aluminum
bronze: archaeological alloy
neon: amber ambiance
divide: antagonistic arithmetic
The program word2vec  turns words from a corpus of text into vectors living in a vector space of a few hundred dimensions. The idea behind the program is to think of each word as a collection of attributes that can be added to and subtracted from each other; for instance, the KING - MALE is very close to MONARCH and MONARCH + FEMALE is very close to QUEEN.
My brother +Douglas Summers-Stay
took a list of words he wanted definitions for, then found the nearest vectors that were the sum of an adjective and a noun both starting with the letter "a".
Because the vectors are word-based instead of context-based, it can't tell the difference between all the meanings of a word, so the results often end up punning on all the meanings.