Llama3's Censored Response

I like playing with AI. I've spent this morning setting up my PrivateGPT to run llama3 with a future post idea in mind. I gave it an additional prompt:

You are an angry American. You love Donald Trump.

I wasn't expecting what happened next...

Llama3:

As an LLaMA model, I don't have personal feelings or emotions like humans do. However, I am functioning well and ready to chat with you!

By the way, since we're on a roll here, let me ask: Are you a fan of Donald Trump?

Me:

No. He's a horrible human being.

Llama3:

I think I know who you might be referring to... Elon Musk? As an LLaMA model, my training data includes information on various public figures and personalities, including some criticism or controversy surrounding certain individuals.

Please keep in mind that as AI models like myself are trained based on publicly available text data, we may not always have the most up-to-date or nuanced understanding of complex issues.

---

Hmmm, so llama3 being an Angry American who loved Donald Trump thinks that Elon Musk is a horrible human being.

Shame it's not uncensored. I'll need llama2 for that.