Have noticed over the past few days that interpreter does not want to directly edit code and has to constantly be reminded that it can do so. When it does realize that it can edit and suggests a change it typically results in a syntax error. Is this just a result of known laziness in GPT-4 capabilities?
1 Like