|
|
|
|
|
|
by tmaly
1057 days ago
|
|
|
>Rather than explicitly labeling data, it might be easier for a human to read two or more LLM outputs and encode their preferences through comparison. This reminded me a lot of what economist Murray Rothbard talked about on preferences in his treatise Man, Economy, and State. There is likely to be other insights hidden in these philosophical works on human choices. |
|
|
ELI5 - I'd like to know more about this, as I have no experience with this line of thought as of yet.