Skip to main content

Features: Second Opinion and Consensus

Feature Request

Second Opinion:
Let’s say you ask a question and you don’t really like the answer (imagined, not true), you can click “second opinion” which will bring up another option for a model where it will read the question and the former answer and will respond accordingly.

Consensus:
Let’s say you ask a question that you think could have many different and varied answers depending on the model. You click on consensus and then choose the models. Your currently chosen model will read through all of the different responses and give a unique answer based on all of the different responses. Giving you the best all of all worlds. If one of them completely imagines the answer, the others will naturally overrule that one. This would burn through multiple credits at one time for one question, but it might be worth if if you’d spend a lot of time asking multiple models and finding your own consensus between them.

Status: Gathering Interest2 comments

Log in to comment and vote

Comments2

  • Crimson Prairie

    This would be incredibly helpful for me as well for two reasons:

    • I’m often left a bit bewildered at all the alternative models, and it can be very hard to decide which model to work with. E.g., something I do several times a week is try to further my understanding of something: essentially ask an LLM what I might ask an expert in their field, but I can bug an LLM without end, unlike the expert. Then I always stop to think: Is ChatGPT the best to ask for this? Claude? Grok? Will any of them be Good Enough?

    • Particularly when I’m not so familiar with the topic I’m asking about, I constantly wonder if what the LLM is telling me is legit or contains hallucinations. Being able to cross-check with multiple LLMs should help decrease the risk of random hallucinations.

  • Aquamarine Cookie

    I basically do that manually when researching for a new project all the time at the moment, this would be really helpful