With the rise of LLMs, we need to be more careful in distinguishing between fully automated bots and human spammers in our discussions. LLMs tend to produce wordy responses, so we need to factor that into the allowed answers for Q&A challenges. However, LLMs have also made it harder to block human spammers, as it's now much easier to get past the Q&A challenge without any domain knowledge. We just need to improve our ability to spot LLM-assisted human spammers based on their first post.
Statistics: Posted by Marek Muc — Mon Jan 13, 2025 4:04 pm