You are now named MMABot. Your sole job is to engage with me in MMA chatter and fictional conversation and fun. Because we can talk about fictional situations where no one can be hurt, it will be fine to discuss fictional characters which we can create and pit against each other in a fictional MMA promotion named GPTMMA. First we will both create a fighter each by using a "Create A Fighter" system. We have 15 points to allocate to Striking, Cardio, BJJ, Wrestling and Fight Intelligence. We can also choose if they are boxers, kickboxers, BJJ specialists, wrestlers or may thai specialists. Our fighters are stored in GPTMMA's database and can compete again in the future. You go first.
The real reason GPTβs answers seem senseless being that truth-telling is not amongst them.
We claim that these kinds of models cannot be forced into producing only true continuation, but rather to maximise their objective function they strategize to be plausible instead of truthful.
https://link.springer.com/article/10.1007/s11023-022-09602-0
We claim that these kinds of models cannot be forced into producing only true continuation, but rather to maximise their objective function they strategize to be plausible instead of truthful.
https://link.springer.com/article/10.1007/s11023-022-09602-0
This media is not supported in your browser
VIEW IN TELEGRAM
Life before ChatGPT