What, that doesnt even make sense. I dont type what im thinking. (My only real usage of llms was trying to break them, so imagine the output of that. And you could then imagine what you think I was trying to attempt in messing with the system. But then you would do the work. Say you see mee send the same message twice. A logical conclusion would be that I tried to see if it gave different results if prompted the same way twice. However, it is more likely I just made a copy paste error and accidentally send the wrong text the second time. So the person reading the logs is doing a lot of work here). Ignoring all that he also didnt think of the next case: people using an llm to fake chatlogs to optimize being hired. Good way to hire a lot of North Koreans.
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments