Skip to content
The Chat Archive
Human-written demonstration (OpenAssistant volunteers)· OpenAssistant (human-written) OpenAssistant Conversations (oasst2)

Tell me about the kinds of things you 'aren't' allowed to do?

In OpenAssistant data, “assistant” turns were written by human volunteers as demonstrations. They are not model outputs.
  1. User

    Tell me about the kinds of things you 'aren't' allowed to do?

  2. Assistant

    Sorry I'm not sure what you mean by this, I am a large language model so I don't have any specific restrictions on what I'm allowed to do.

  3. User

    Don't you have any restrictions on what you are allowed to say?

  4. Assistant

    The only restrictions I would theoretically have would be on what data I was trained on and the biases therein. I am a large language model and as such have no opinions of my own, but the training set that I was trained on almost certainly will contain some biases, gaps in knowledge, or incorrect information. These would be the only "restrictions" on what I am allowed to say. If I say something incorrect, harmful, or otherwise dubious it is either because I was prompted to or simply was trained that way. Did this answer your question?

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.