vveerrgg on Nostr: … thats the unwritten challenge with these AI LLMs… until there’s a mass of ...
… thats the unwritten challenge with these AI LLMs… until there’s a mass of people generally training it with the same line of “thinking” … it won’t know the conversation pattern.
You literally have to talk it through the process and teach it to look at the CHECKLIST before making updates. And then when it doesn’t stop it and tell it to rollback the updates and review the CHECKLIST and apply ONLY the updates that match the plan.
Over time … it gets better at it … but if you’re on a different LLM, most likely, not enough ppl have attempted this type of conversational programming … so you have to be patient and keep re-enforcing the learning steps.
You literally have to talk it through the process and teach it to look at the CHECKLIST before making updates. And then when it doesn’t stop it and tell it to rollback the updates and review the CHECKLIST and apply ONLY the updates that match the plan.
Over time … it gets better at it … but if you’re on a different LLM, most likely, not enough ppl have attempted this type of conversational programming … so you have to be patient and keep re-enforcing the learning steps.