According to my own experience dealing with/learning auto-gpt for the last 4-5 weeks, it all comes down to context. If the STM doesn't contain enough context it doesn't matter what you tell it to do. It will try one command after another blindly even if you tell if that this and that command didn't work. If however enough context is retained in STM autogpt can figure out quite quickly what it did wrong and find the correct way forward. #2
Description
According to my own experience dealing with/learning auto-gpt for the last 4-5 weeks, it all comes down to context. If the STM doesn't contain enough context it doesn't matter what you tell it to do. It will try one command after another blindly even if you tell if that this and that command didn't work. If however enough context is retained in STM autogpt can figure out quite quickly what it did wrong and find the correct way forward.
Originally posted by @W3Warp in Significant-Gravitas/AutoGPT#2987 (comment)