Raw LLM Responses

Inspect the exact model output for any coded comment.

Comment
It's pretty easy to give them long-term goals, though. Just let them write to their own initial prompt. I have this running on my desk. EDIT: why downvotes? I just gave the GPT-4 API a function it could call to write a string to its own initial prompt, allowing it to adjust its basic protocol. It’s not rocket science. It means it has persistence between sessions and can store goals.
reddit AI Jobs 1685880878.0 ♥ -7
Coding Result
DimensionValue
Responsibilitynone
Reasoningunclear
Policyunclear
Emotionapproval
Coded at2026-04-25T08:33:43.502452
Raw LLM Response
[ {"id":"rdc_jmubako","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"}, {"id":"rdc_jmursi5","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"approval"}, {"id":"rdc_jmu86ru","responsibility":"none","reasoning":"deontological","policy":"none","emotion":"outrage"}, {"id":"rdc_jmut5gw","responsibility":"none","reasoning":"deontological","policy":"none","emotion":"outrage"}, {"id":"rdc_jmv8cms","responsibility":"none","reasoning":"deontological","policy":"none","emotion":"mixed"} ]