nature = the model
nurture = the prompt

various methods exist for improving reasoning, but they all evolve around adjusting one or the other

@operator & /operator | 💭 about BCI, bioinformatics, search, chaos, decentralized AI | join me in /entropybender /llm

Product researcher & /ux strategist, building a decentralized Farcaster incubator with /concept-testing, journaling /angeluser | nf.td/markfishman

conversation then adapt my prompting based on what the results are like

but often for most tasks it's a chore to adapt the prompt so i don't rly do it

this may count as neglect

Do you prefer to give a perfect prompt or have a conversation with models?