3 ms·

I could probably get any LLM to repeat basically any rhetoric with enough prompting, you're right. The problem is that Grok explicitly leaned towards that with

I could probably get any LLM to repeat basically any rhetoric with enough prompting, you're right. The problem is that Grok explicitly leaned towards that with shockingly little prompting.

acquiesce 1y ago
Isn’t that what mostly happened with Tay though?
- spookie 1y ago
  Tbh it is being trained off data of the same platform.