Which of the following would you find most undesirable in a widely-used AI tool?
	
	A. Always report the most common views, hiding uncommon opinions entirely.
	B. Be less effective/accurate with uncommon/minority situations.
	C. Mimic the biases and bigotries of humans as present in training data.
	D. Provide bad actors with tools to further their nefarious goals.
	E. Show each user what they prefer to see, reinforcing whatever they believe.



















	Which of the following is an example of what we mean in computing by "bias"?
	
	A. Exaggerating information, making big numbers bigger and small numbers smaller.
	B. Only hiring people who resemble me in gender, skin tone, attire, and accent.
	C. Rolling dice instead of doing computation when asked a question.