Let’s talk about AI slop.
Slop has been slapped onto “poor” AI output, but it’s actually not unique to LLMs. Human beings have been putting out slop forever, we’ve just called it different things: “cliche,” “cheesy,” “dumb,” etc.
Slop, at its core, is a low-hanging fruit idea. Something that lacks novelty and intellectual precision. Something that many have said before and because of that instead of tickling an intellectual itch, it’s more of a scratch to the brain of the recipient.
And what makes an idea low hanging? It’s a human or AI output that is the highest probability pick of an idea. And because it’s the highest probability, it’s the most expected. And because it’s the most expected it’s the least original. Hence the scratch vs the tickle.
Has anyone encountered a good definition of “slop”. In a quantitative, measurable sense. My brain has an intuitive “slop index” I can ~reliably estimate, but I’m not sure how to define it. I have some bad ideas that involve the use of LLM miniseries and thinking token budgets.