The first line below (in italics) is the prompt I gave to OpenAI's GPT-3. Along the way, I made sure to re-prompt GPT-3 by prepending each run with "Philosopher AI:". This helps to keep it from rambling, and start a fresh thought.
After about the first ten quotes, we started appending a few words to the prompt to get more specific answers. All generated content is in bold, to distinguish it from the prompts.
Most of these are first tries. I removed one due to repetition, and one due to being off-topic. Otherwise success rate has been incredibly high (>90%).